THE BOOK--Playing The Percentages In Baseball

Pineiro, 2/13, paying for 1.7 WAR.

Pineiro had a horrible forecast coming into the 2007 season, half-way between a replacement-level pitcher and a starter.

For the 2007 season, it’s was a strange one.  He looks like he performed better as a starter than as a reliever, but that’s because he did exceptionally well with men on base.  This is a talent that few hold.  A better sign is his OBP/SLG in his two roles.  As a reliever, he was horrible, basically replacement-level, or worse.  As a starter however, he was a bit below league average.

The expectation is that your performance as a reliever should be no worse than that as a starter, and likely significantly better.  Pineiro showed a reverse split, which may be due to sample size, or a change in environment (Bos/STL), or he really hated being a reliever.

This is one of those cases where you’d want to look at the pitch-by-pitch data.  Barring that, we’re stuck.  If Pineiro can provide say a .450 record as a starter, with about 15 full starts (say around 135 IP), that makes him a 1 WAR pitcher.  For him to go all the way to a 1.7 WAR, which is what the Cards are paying him, he’d have to be a .480 pitcher with 17 full starts.

I don’t think that is justifiable, so I’ll have to say that the Cards overpaid for him.

That said, if you are going to overpay for someone, at least sticking to 2 years is palatable.

If ever there’s a case of needing to sign someone to an incentive-filled deal, this is it.

Tango, this is kind of related and kind of not.  I cannot for the life of me figure out why Theo signed Pineiro to a $4m deal this year.  Do you think it was a matter of competition for his services necessitating that price?  Or is it some kind of baseball rule having to do with years of service, arbitration, or something like that?  I just can’t believe that some other team was willing to pay even $3 million, so I wonder if this is an issue of Theo somehow misreading the market.  Seems trivial, but this is one move the Sox have made that I just can’t follow.  If you want to take a chance on a guy, fine.  But couldn’t they have done it for $1.5 million or something?

I think it was just a dumb signing, same situation as Bill Stoneman signing Shea Hillenbrand for 6 million. Or a much more costly mistake like giving Carlos Lee 100 million - which the Astros probably don’t even realize was a mistake since he got his 30 hr and 100 rbi.

Piniero was a free agent and any team that signs him is under no obligation to pay him more than the league minimum.  If service time had anything to do with it, Julio Franco would be getting A-Rod money.

Apparently Piniero has a very good agent, who worked his magic on Epstein last year and gets an even better deal from the Cards.

Coming into the 2007 season, the Bill James forecast was for a slightly below average pitcher.  Marcel had him much closer to replacement.  Chone, for some reason had him forecast for 60 K, 25 BB, plus a 4.45 ERA, which was lower than any of his 2004-2006 ERAs.  Clearly, Chone must be wrong in its forecast.  And yet, he ended up 60/26, 4.33 for the 2007 season!  Hopefully, Chone forecast him for a reliever (85IP forecast), so that would explain the very favorable forecast.  It’s just interesting that he stunk as a reliever.

Anyway, he should have been signed last year to a 1/2.5 or so deal.  Redsox have the luxury to overpay for potential.

Yes, I did a starter to reliever conversion on him.  At the time I ran the projections, he was supposed to be the new Red Sox closer with Papelbon going to the rotation.

Do NOT forget that there is a gigantic difference in the NL and AL right now.  A pitcher going from the AL to the NL “gains” something like .3 runs in ERA.  (Same is true for hitters.)

If you are an NL team you MUST get almost all of your players from the AL, as they will likely be undervalued as compared to NL players.

For those of you who don’t know what I mean…

For example, if a starter in the AL has a true talent ERA (whatever that actual number happens to be - doesn’t matter whether it is 4.3 or 4.8) of exactly league average (so he is a league average starter - obviously) and he goes to the NL, he will have a true talent ERA of around .3 runs LESS than the NL league average (again, that could be 3.5 or 4.5).

That has nothing to do with the parks or the hitting.  It is because the average pitcher in the AL is .3 runs in ERA better than the average pitcher in the NL (if they played in the same park against the same batters).

Good stuff Rally. 

A .500 reliever (+.030 above replacement) for 85 IP (almost 9 full games), still registers as barely above replacement (+0.3 wins).  That would make his salary offer from the Redsox at $1 million.  How you can (a) go to free agency and (b) find a closer for $1 million is clearly ludicrous unless you (c) offer him $4 million to do so.  The Expos did something similar when they signed Graeme Lloyd and Hideki Irabu: overpay for an asset so that it looks valuable.  That would explain Carlos Lee.

MGL, I get pretty much the same results for league difference.  I ran a dead average pitcher through the CHONE system, and I get a 0.42 difference between pitching in an average AL park vs NL.  The American League has about a .10 higher ERA the last few years, so relatively, the AL guy is .30 runs better.

For Piniero I have a 4.53 ERA in St Louis, assuming average defense.  He would be 5.17 in Boston.  The projection is split with 19 games starting and 37 games total - I’ll add G and GS this year so people know what role I’m projecting.

I haven’t gotten to projecting team defense yet but St Louis will probably be above average, mainly because they have Rolen and Pujols.

Of my top 30 projected starters, 21 are in the American League.  The disparity is really apparent when you watch these playoffs, and for 3 of the NL teams, they have very good aces (Hamels, Webb, Francis) but when you get to the #2 guy you ask yourself “how did this guy get to make a playoff start.  Of course the Cubs had three decent starters and still got swept.  Funny game.

http://www.sabernomics.com/sabernomics/index.php/2007/10/cards-sign-joel-pineiro

JC has Pineiro’s value for performances in 2005/06 as totalling 16MM.  I think there’s definitely something wrong with how he comes up with that number.  Without running the numbers completely, it’s a safe bet that Pineiro was one of the worst starters over those two years.

Ah, what the heck:
http://www.bb-ref.com/pi/shareit/m4z2

Of all starters with at least 300 IP in 2005/06, 86 starters qualified.  Pineiro had the second worst OPS+ of the group.  (I forgot about you, Ramon Ortiz.) Pineiro also had the worst ERA+ of the bunch.

Even dropping the qualifier down to 200 IP, he’s 136th out of 140 in ERA+.

http://www.baseballprospectus.com/dt/pineijo01.php

BP has him at -52 runs relative to average over 355 IP, or -1.32 runs per 9 IP.

That’s below replacement level performance.

Sorry, JC.  The numbers don’t add up.

I have updated the salary chart for this off-season:
http://www.tangotiger.net/salary.html

The 2007 and 2008 charts are both available.  I went with 4.4MM per win for the 2008 chart, which is 10% higher than the 4.0 for the 2007.

Obviously, better research would give us a better number.  I’m happy with the “rule of 10” for now.

There’s also a link back to this blog from the page, so that should make life easier for all.

I’m just waiting for BDB to get their data up-to-date so that I can run Marcels, and we’re off to the races…

Good stuff.  It would be nice to know the marginal dollar value of a marginal win once and for all, but I think your assumption is going to be close either way.  It’s funny how just a few years ago we were saying $2 mil per win is what you should strive for and now we’ve doubled that number.

With that said, because your number is for free agents, would you use that same $4.4 mil per marginal win to calculate the value of an arbitration/non FA player?  Wouldn’t be (technically) have to create another $/marginal win number for these players that are not FA or you have already signed (and what to know what they are now worth in present value)?

Going further, if we had access to team financial data and could calculate the dollar value of a marginal win for each team, I guess that would take care of non FA.  I really wish someone could get hold of that data because that combined with slwts would get use nearly to 100% of what each player is really worth.  I know a lot of players’ value wouldn’t change that much, but I think the information would be worthwhile.

You’re probably new around here!  I use the following:
20% 2.x to 3 years less a day
40% 3 to 4 years less a day
60% 4 to 5 years less a day
80% 5 to 6 years less a day

For guys who had exactly 3 years of prior ML service, their arbitration value is 40% of the free agent value, or 1.76MM per win.

The 2.x means only the “super twos”, or guys who qualify for arbitration.

I was also thinking that there’s a “lag”, since arbitration awards are based on prior comps.  But, for multi-year settlements (like Chase Utley), there wouldn’t be a lag.

Still, I’m going to be doing work on these classes of players, now that two people so generously gave me the long-needed ML service time data.

The 2007 multi-year arb players were very low (Utley, Mauer, McCann), which makes me think the 20/40/60/80 is too generous (or more likely that these players got scr-wed).

Another thing is that there’s greater revenue sharing (meaning fewer marginal dollars).  Think about the NFL, where a substantial portion of their revenue is shared.  If all their stadium revenue was also shared across all 30 teams, then guess what: there is zero monetary incentive to get a better product, and therefore, the marginal dollars per win would be zero.

So, it’s possible that the 10% growth I’m giving should be much lower, especially if the MLB revenue growth is due to MLBAM.

However, I feel quite confident that MLB owners have no fiscal responsibility, and are pure Picasso purchasers: they will spend to their budget, even if market forces allow them to spend far below that.  They are like kids at a toy store, thinking this is the only toy store around, and all the other kids will soon gobble up all the toys.  They end up creating a market that would otherwise not need to exist if they simply applied a modicum of rationality.

Greg Maddux: signed for 1/10, paying for 2.3 WAR.

I don’t have my Marcels, and since this is just a 1-yr deal, I’ll just wing it.  He will probably be forecast for 180 IP for 2008, meaning 20 full games.  He did have a 4.14 ERA, and his FIP was a fanstatic 3.52.  But, this is Petco.  The impact of parks on runs is listed here:
http://www.insidethebook.com/ee/index.php/site/comments/run_impact_in_parks/

We need to bump up Padres’ ERAs by 0.50, which is fairly substantial.  This basically makes Maddux a .500 pitcher.

Since a replacement pitcher (as a starter) is a .380 pitcher, that makes Maddux +.12 wins above replacement per game, or +2.4 wins above replacement.

This is a huge bingo.

(Maddux, I would bet, has figured out that he can keep “looking” good by pitching in a pitcher’s park.)

Schilling: 1/8, plus 2MM “healthy” bonus, 3MM “performance” bonus, 1 MM for any Cy Young votes paying for ??.

Let’s assume that he’ll hit 1.5MM on the healthy bonuses, and 2MM on the performance, and won’t receive any Cy votes.  That makes it a 1/11.5 deal, paying for 2.6 WAR.

Let’s presume 162 IP (18 full games), as a slightly above average pitcher (say .525).  That makes him +.145 WAR per game, or a total of WAR of 2.6.

I swear that I wrote this on the fly, and in just 5 minutes.

A huge bingo.

I just thought of something.  The AL is the superior league, as we’ve shown say an 8 win gap between the average AL and average NL teams.  But, the average AL team won 82 games and the average NL was 80 games.  There’s an additional 6 win gap that needs to be applied (+3 wins for each AL team, and -3 wins for each NL team).

+3 wins means about +1.3 wins for the pitchers and +1.7 wins for the nonpitchers.  If you pitch 162 innings (18 full games), that’s 1/9 of 1.3 that needs to be adjusted, or an extra 0.14 wins to be added to Schilling, and 0.16 wins to be subtracted to Maddux.

So, when I said that Schilling is a 2.6 win pitcher, he’s probably a 2.74 win pitcher.

And calling Maddux a 2.4 should really be a 2.24.

Not much of a difference obviously, especially since I’m just doing this with the least amount of effort.

Here’s the free agent tracker at ESPN:
http://sports.espn.go.com/mlb/features/freeagents

***

Mike Lowell: looks like he’s a +1 win hitter, and Fans still see him as a +1 win fielder.  Dewan’s Plus/Minus concurs somewhat.  So, he’s +2 wins above average.  Take off 0.5 wins for aging.  No adjustments for 3B.  Playing time expected to be full-time.  So, +3.5 WAR player.

I expect him to sign for: 2/30, 3/43, 4/55, 5/65 deal.  What’s the word on the street?

Omar Vizquel:

This is where more research on aging would be nice.  He’ll be 41 years old, and the Marcel forecast of 5/4/3 just doesn’t seem to me that it should apply.  I have to believe that the further away you are from age 27, then the more weight you have to give the recent season.  Something like 5/2/1, or something.  From 2004-2006, Vizquel was a shade below league average hitter (not for a SS, but for all hitters).  But in 2007, he was one of, if not the, worst hitter in the league.  (Or, I should accurately say “worst performing...”, which is a big difference.)

Furthermore, there’s a regression toward the mean component, which is league average.  But, he is 41 years old.  And, he should be regressed toward a different population mean: that of 41 year olds.  The outlook for that is not good. 

The optimistic forecast is to call him a -1.5 win as a hitter.  But, given what I said about forecasting an older hitter with heavier weight to his recent performance (disasterous for Vizquel), and giving an older hitter an even sharper decline, I think a -2.0 wins as a hitter would be appropriate.

On the other hand, his fielding is still stellar!  Not only do the Fans have him as one of the best in the league (+2 wins!), but Dewan’s Plus/Minus has him very high as well.  But, a 41-yr old SS has got to age faster than we’d allow.  Don’t forget, an IF relies on his legs alot, and speed is the fastest declining tool to a player.  And Fans are notoriously biased with aging stars.  Calling him a +0.5 win SS is as low as I could go, fielding-wise.  +1.0 might be more reasonable.  Somewhere in there.

SS also get a +0.5 win premium just for being SS. 

So, his WAR over 162 games would be something like +1 win, maybe +1.5. 

Again, at his age, how much playing time can we expect?  130 games means 80% of the season, so that puts him at +0.8 to +1.2 wins.

So, my take is that he would deserve a 1/4.4 deal.

It may be worthwhile to put this discussion topic at the top of the page as long as the off-season goes. Like the postseason strategy thread during the postseason.

Tango, I’m correct in assuming that you’re supposed to try to find the players true talent level before using your scale, right? It looks like that’s what you’re doing, but I just wanted to make sure (I’ve already used it a couple of times on my blog/message boards ... thanks for it, as usual!)

It looks like to me the Schilling and Maddux deals are going to be fine ones, especially for pitchers.

The suspected Vizquel deal took a lot of flak over at BTF as I was reading through that thread, but as you show here, I don’t think it’s as bad as some people think (mostly due to Vizquel’s still very good defense).

Well, as a general rule, I don’t want to pay free agent prices, since all those players are paid double what they should be paid.

But, if you HAVE to go the free agent route, that’s the going rate.

Now, you don’t HAVE to go get a free agent who is a 1 WAR player.  But, teams still seem to pay those prices.  Go figure.

And yes, you are paying for true talent, always and forever.

23/Tango: Thanks.

So, basically if a deal lines up with your chart, it’s in line with what the market is paying for free agents. I think I got it…

And I agree with you on your second point there, of course. I’m sure many teams overlook younger/cheaper players in their organization to pay a “proven” free agent. Even if they get a fair deal on the market, it doesn’t always mean that it’s a good decision.

Please, please don’t forget that if a team signs a player at his market rate according to Tango’s numbers, they are exactly as “smart” as the average major league team, which ain’t too smart.  Before you start congratulating a team on a signing that is around the proper value, keep that in mind.  Not to say that a team should never sign a player at or above this value.  That is not what a team should be looking for anyway.  The ONLY thing a team should care about is how much revenue they will draw compared to how much they spend, and as we know, that is complicated.  Very complicated.

But, as a rule of thumb, any smart team (smarter than the average team) should look to be signing FA, on average, for quite a bit less than 4.4m per win.

So, as a general rule, if a team signs a player for around what they SHOULD get paid according to the “4.4 per rule”, you are entitled to say, “O.K., they are just as smart or dumb as the average team.” Before you say that something is a “good signing” wait for it to be well under market value!  This is not like most businesses where market value is the proper baseline. Not even close. In baseball, market value is the sum total of 5 really smart teams, 10 slightly smart ones, 10 dumb ones, and 5 really dumb ones.  If you can’t beat market value (on the average) for FA signings, you ain’t a sabermetric team.  Don’t forget that for every overpaid player (one who signs for 6 or 7 mil per), and there will be many, there HAS TO BE equivalently underpaid ones.  The smarter teams are likley to gobble up these underpaid players.  As well, these smart teams are not likely to overpay (much) for anyone, especially if they are not large market (high revenue) smart teams, like the BoSox (who can pay anything for any player as long as that player will give them at least as muich marginal revenue as they pay for him).

Marginal revenue… or blood.

"I expect him to sign for: 2/30, 3/43, 4/55, 5/65 deal.  What’s the word on the street?”

There haven’t been a lot of specific numbers associated with Lowell.  An early rumor was that NYY was willing to go 5yrs/70M which scared Sox fans, but is pretty close to your 5 year number.

Sox have said that they would go 3 guaranteed years and Lowell has reportedly indicated he was open to a hometown discount.  My initial feeling was that the Sox would max out at 3/42, but some other team would get him for 4/56.  Recent reports have speculated that the Lowell market is a bit soft, however.

Honestly those were my numbers before I saw yours.  So from what I can tell it looks like the rumored Lowell market is surprisingly rational for an aging WS MVP with great clubhouse intangibles.

Tango, I forgot where you get your 4.4 per from, but in any case, don’t teams spend considerably more for pitching wins than batting wins (as you would expect them to, as they overrate pitching as compared to offense), in which case, they must be evaluated separately?

philly: good stuff.

MGL: I used 4MM last year, and simply bumped it up by 10%.  I based it on studes’ work on Net Win Shares Value, which is an annual must read.  (Where is studes anyway?)

I don’t think they necessarily spend more for pitching (esp with the repl level I use: .380 for starters, and .470 for relievers).  It seemed to work out fairly ok last year, with all those pretty good pitchers.

Tango/14.

My bad.  We’ve actually emailed each other a few times.  I definitely remember you talking about arbitration value before, but I forgot about the numbers.  Thanks for posting them.

While we use the $4 or $4.4 million per win value for FA now and the 20/40/60/80 arbitration values, if we were able to calculate the exact value of a marginal win for each team, would the former two numbers used to calculate a player’s dollar value still be necessary?  Thinking out-loud here, I guess if we knew the exact value of a marginal win for each team, then we could really evaluate each contract signing as “good” or “bad,” because as MGL pointed out, the $4-4.4 mil/win value is merely what the average market price is nowadays, not the optimal price each team should/can be spending.  Just something to think about I guess…

I like these threads because the 4.4 per win is from the perspective of a fan.

If you look at marginal revenue you’ll probably get a figure around 2 millon.  Maybe a bit more since that was studied awhile ago, but certainly not 4 million.  A team that looked to profit maximize would refuse to pay market rate for free agents, miss out on almost all of them, and lose a lot of games.  But at least they’d make money.

I’m glad I don’t root for one of the teams that takes that approach.  Teams can afford to spend 4 to 4.5 million per win, and most of them do. I’m happy when my team buys a good expensive player at 4.4 per win, because it means they will win more, though it may not be great for their bottom line.

When a team overpays, like in the Carlos Lee example, fans should be upset, because the team is using its resources inefficiently and will not win as many games because of it.

Juan Uribe: signed for 1/4.5, paying for 1 WAR.

Both my sources list his service time as 6y, 89 days, meaning he would have otherwise qualified for free agency, if not for his contract.  The Sox has a 5MM option for him, and they renegotiated down to 4.5MM. 

Uribe is not a good hitter, despite his HR.  He’s a -1.5 win hitter.

The Fans saw him as an average fielding SS last year (0 wins), a substantial drop from his 2005/06 evaluation (+1 win). Dewan’s Plus/Minus concurs with the change, showing a nearly 1 win difference.  However, Dewan had him lower to begin with.  I’ll stick with 0.

He gets the 0.5 win adjustment as a SS.

Aging doesn’t really apply here. 

That puts him at -1 WAA, or +1 WAR.

Basically, a perfect signing.  I was at first surprised that Uribe negotiated himself down, thinking that he must be much better than a 1 WAR player.  But, his fielding must have taken a big tumble last year.

Torii Hunter:

He’s a +1 win as a hitter.  Fans have him as a bit above +1 win as a fielder, while Dewan has him at +0.5 wins.  Fans are also notorious for overvaluing aging stars, though they decked him pretty good between 2005 and 2006.  Let’s give him the +1 as a fielder.  He gets his +0.5 as a CF.  And we drop him 0.5 for his age.  That puts him at +4 WAR.

Expect him to sign for 3/51, 4/65, 5/78, 6/89.

Any word on the street?

A little technical note for nonpitchers.  When I present the WAA, it’s based per 162 GP (or 700 PA).  The replacement level is -2.25 per 162 GP (which is the same as -2 per 144 G, or -2 per 620 PA).

So if you have a guy forecasted for 600 PA (86% of a season), and if his WAA is +1 per 162, then his WAR per 162 is +3.25.  And 86% of that is +2.8.

I’ll sometimes fly through and simply do +1 +2 = +3.  I should be a bit more careful here.

Hmmm… I see Torii rejected a 3/45 deal in August, in favor of a longer term deal.  This means the Twins were paying him to be a 3.6 WAR player.  Torii also wants a long-term deal (5 or 6 years).  At the 3.6 WAR level, that means the Twins would have offered 5/65, 6/72 (i.e., according to the Twins implied value, 3/45 is the same as 5/65).  It seems to me that if the Twins opening offer was that close (3.6 WAR, compared to my 4.0 WAR), that they should be able to meet the free agent price.  Unless of course they are in rebuilding mode.

Maybe Torii is looking at being more favorably compared to Wells and Ichiro.

#31, this whole idea of what constitutes a good or bad signing is a complicated one with few definitive answers.  I don’t pretend that I have much insight although I have thought about it a lot (usually to little avail).

The average dollars per marginal win for FA (which is what the 4.4m is supposed to represent for this year including anticipated inflation) is as good as any as a frame of reference or starting point to analyze signings.  On the other hand, it doesn’t mean a whole lot on a team by team and player by player basis.

Not to mention the fact (and I have mentioned this a lot over the years), what do we mean by “good” or “bad” and from whose perspesctive?  For example, the persepctive of a fan who just wants to give his team the best chance of winning the WS or winning as many games as possible (and even that is ignoring the long-term) is COMPLETELY different from the perspective of an owner who wants to make as much money as possible in the short and long term.  And what about the GM who is usually given a set budget to spend and must field and develop the best team possible in the short AND long run as well as keep his job and in some cases, LOOK GOOD to the fans, media, owners, etc. (let’s face it, part of human motivation is “looking good” even if that is not an especially optimal strategy from other persepctives).

Now, there are other considerations, such as the likely non-linear value of the marginal win across the population of players.  Because of supply and demand, the more marginal wins a player’s talent represents, the higher value (in terms of market value strictly) each marginal win.  Let’s say that there were one player worth 5 WAR and all the rest were exactly average (say, 2 WAR).  Well, there would be NO competition among teams for all of those average players; there would be enough to go around.  They would all get paid something less (maybe 75%) than the amount of revenue they provided their team, assuming that all teams/owners were rationale and only in business to make money.

What about that good player (assuming that everyone knows he is the only good player).  Presumably, at least according to the linear model, he would be worth and paid 2.5 times that of the other players (5 WAR versus 2).  But that is not what will happen of course.

In a rational market where every team has the same revenue, that one player’s price will be “bid up” by all 30 teams because he is the only way that one lucky team can be an above average team.  Let’s forget about playoff potential and sweet spots for a second.  What would be be “bid up to” in a rational market?  Not higher than the revenue he produces.  But, somewhat higher than the other players per marginal win.  If the other players were paid 75% of their revenue value, this one premium player would be paid 80% or 85%.

Now let’s throw in the fact that some teams’ marginal wins are worth a lot more than others, which is indeed the case.  All the other players would still be paid the same amount, more or less.  The rich teams would not have to pay any more than any other teams for these players because there are plenty of them to go around. 

But what about this one premium player?  Well, obviously the rich teams can and will afford to pay much more for him per marginal win than for the other players.  And because of the competition among those rich teams, his price will be bid up enormously.

Now, throw into the mix the fact that teams/owners are not rational (from a bottom line perspective).  Many owners don’t care (or they say they do, but they don’t operate like they do, or the people that work for them don’t operate that way and the owner does not know it) how much money they make.  A team, as Tango says, is like a Picasso to them.  And the more they win, the prettier and more prestigous that painting is to them.  What happens then?  Well, that premium player is going to get bid up even more as teams who cannot really “afford” him will enter into the mix, and teams that WANT him no matter what will actually bid more than they can “afford” as well.

That is why premium FA’s are worth more per WAR than non-premium ones.  Simply because there are fewer of them and the law of supply and demand is part of the valuing equation.  Basically the valuing equation for a marginal win is: value in marginal dollars plus the effect of supply and demand.  If you think that the first thing takes care of the second, you are wrong as my example clearly illustrates.

There are many other issues and complications of this “is a signing worth it or not” equation.  I will comment further, not the least of which is that each team (at least each category of team in terms of revenue) is COMPLETELY different in terms of evaluating FA contracts.

I mean it is mildly interesting (and that is a stretch) to evaluate a player signing in terms of his projected WAR (you can use any one of a number of equally good forecasting systems) and then see how much he “should” get paid according to a “4.4m per” standard.  That is going to get old quickly.

Boston paper today reports Sox have a 3 year $12M to $15M deal on the table for Lowel - or 3 year $36M to $45M. The high end being very close to your 3/43 Tango.

Cool, thanks.  I see in the paper they qualify it as a “strong offer”; what the h-ck does that mean??

***

http://sportsillustrated.cnn.com/2007/writers/jon_heyman/11/08/scoop.thursday/?eref=mostpop

Jon Heyman is doing his free agent analysis, too.  Let me compare.  He called for ARod for 8/256, and now is saying 10/320.  I said 8/252 (or 10/306).  Looks like we both think the same thing here.

He says Torii 6/90.  I said 6/89.  Awfully close!

He says Lowell 4/52.  I said 4/55.  Another tight one.

I didn’t look at his numbers for the other players I haven’t done yet.  Not that this would bias me anyway, since I spend all of 5 minutes on each player, and what I do is fairly verifiable.

With many evolutions, you wake up one morning and notice that everything has changed. That is not necessarily due to a “tipping point” but sometimes it is.  Sometimes it is just the illusion of a tipping point.  Like when you lose your hair a little at a time and you wake up one day and notice that you are bald!

Anyway, I think we may have reached the point where teams and even the media know pretty much how to evaluate and pay players.  Not completely, but alot more than just a few years ago.

Either that, or this guy and others are just copying Tango’s numbers…

The 2008 forecasts from Bill James (BIS) are up:
http://www.fangraphs.com/statss.aspx?playerid=841&position=C

***

Jorge Posada:

Bill James has him forecast for his career average, which makes him a +2 win hitter (per 162 games).  I don’t have my Marcels for him (or anyone) yet, but that’s probably what I’d have, prior to aging.  Fielding-wise, probably an average catcher.  Give him the +1 win bonus for being a catcher, and -0.5 wins for aging, and he’s +2.5 WAA (per 162).  Add in the 2.25 for replacement, per 162, then reduce the total (4.75) by 25% for playing time, and he’s a +3.5 WAR player going into 2008.

That puts him in the same salary class as Mike Lowell: 3/43, 4/55.

Let me look at Heyman..... 4/52.

Yeesh.  What’s the point!  I think MGL is right…

Interestingly enough, Cashman has stated he will sweeten the 3 year $40M deal currently on the table - Perhaps to 4 years $55M. Tipping point indeed

http://sportsillustrated.cnn.com/2007/baseball/mlb/11/08/yankees.posada/

Omar signed for 1/5.3, compared to my expectation of 1/4.4.  His implied value of 1.2 WAR, which is a bit on the optimistic side.  Fair signing.

Pedro Feliz:

His fielding is sensational, Gold-Glove caliber.  And I say this without ever having seen him (or noticing him).  I’m no expert, which is why I rely on Giants fans.  Fans call him +1.5 wins as a fielder.  John Dewan has him as the top 3B from 2005-2007 (+61 runs).  Clearly, he’s fantastic with the glove.

His hitting on the other hand is at the -1.0 or -1.5 win level.

Overall, he’s somewhere between a league average 3B to a somewhat above.  Add in the aging, and he’s at league average, or a bit below.

That puts him right around a +2 WAR player.

I’d sign him for 2/16, or 3/21.

My guess is that good fielding, no-hitting 3B are undervalued.  Let’s see who goes for him.

I agree that he and his ilk are likely to be undervalued.

Let’s play this game with John McDonald. Signed for $3.8 for 2008/2009. He became really, for the first time in his career, the everyday regular at short in June. By several measures he appeared to have an outstanding defensive season in 2007, UZR had him at +21/150, PMR at +21, the Fielding Bible gave us 3 year data sample, so I used it, extrapolated to 1440 innings from 2005-2007 he comes in at +20.7 runs. The Fans had him as the second best shortstop in the game behind Everett. He’s a hard guy to get a handle on historically because of previously being a utility/spot starter, the sample size is not great. Still I think there’s enough data out there to call him a 2 win defensive player ( I’m not sure how much age will effect his performance moving forward, especially considering how Omar has aged, a 2-3% decline per year?).

Hitting wise he’s abysmal. Using B.R’s base runs and factoring him up to 650 plate appearances, he’s -35.3 runs compared to the league. I actually took all A.L. starting shortstops, factored them to 650 P.A’s and determined the average production to be -14.5 Base runs for the shortstops. So he’s basically a minus 2 win hitter at his position.

This is where things get murky for me. Going forward if he’s a full time player he’s a - 2 WAA hitter, given aging he’s a + 1.5 WAA defender, this takes him to +1.5 WAR ( I won’t add the SS premium since I’ve factored this in all ready by comparing his offence to the average SS) (Is this right?)

Basically he’s performing at the level of Omar (+1 .5 WAR). If he played everyday you’d pay such a player $ 6.75 million. Now he’s never been a regular, and it remains to be seen if he can maintain his 2007 fielding at the 2007 level over the course of a full year. Spending $1.9 million/year for this level of performance, even as a utility player seems good bang for the buck.

I think you are giving him too much defensive credit.  Even using your numbers (Dewan), you have to regress and keep in mind that only the best SS in the prime of their defensive careers (young) are in the +20 (2 win) range.

Plus, my numbers for him are not even close.  He was +14 (per 150 games) this year, but in prior years, he was minus.  The fans will tend to rate a player on how he has done lately if he is new at a position (and on his repuation if he is old at a position), so I would not put TOO much credence in the fan rating for him.

I would put him at 1 win, 1.5 win tops, and that is using Dewan’s data and not mine.  Using mine, I would have to put him at 1 win tops.

Plus, figure 2 runs a year loss at SS with age.

Hitting-wise, you are probably right on the money, maybe a little too pessimistic because you are not considering regression toward the mean since he does not have that many ML PA.

He is, as you say, an awful hitter.  Truly awful.  He has been (in actual performance) well below replacement level even for a SS.

However, with regression, I would have to put him at 1.5 wins below average or around .5 above replacement, at best.  Considering that I don’t quite use 2 wins below average for replacement, I am going to comfortably call him a replacement hitter at SS.

So, that makes him 1.5 wins above replacement, but I think it is more like 1 win, using my defensive assessment.

I agree that it is probably a decent deal.  The fact that it is not widely known that he is a very good defensive player and his abysmal hitting, probably makes him an undervalued player.  Who, other than a team that “knows him,” would want to pay this guy anything at all?

Then again, since there is fine line between replacement and a win above replacement, and there are a lot of players in that category, you are going to find a lot of “bargians” in that area.

Plus, it is debatable that a .5 win (above repalcement) player is actually worth 2 or 3 mil, as Tango’s formula would suggest.  Remember the issue of supply and demand and the non-linear relationship between marginal wins and value among players I mentioned in a previous post?

#47, Exactly.

I think the formula works very well for players average and better.  For players between replacement and average, they’ll get paid the 2-3 million the formula says they should if they are well liked, but if you try to save money in that area you can probably get a player just as good very cheap.

You can buy a billion players that are between average and replacement, but no number of such players will win you 81 games.

Here’s my take on this year’s Japanese prize, Kosuke Fukudome:

http://lanaheimangelfan.blogspot.com/2007/11/kosuke-fukudome.html#links

Nate Silver at BPro developed a statistic called MORP (market value over replacement player) about two years ago. It applied exponential rather than linear growth to a player’s market value. The formula is:

(WARP^2 * $212,730) + (WARP * 402,530)

The issue with the stat, of course, is that WARP is a lousy stat with its bogus handling of replacement level and the limitations of the fielding component in it. WARP saw John McDonald as a 4.3 WAR player largely because it sees a +20 defender as a +40 defender based on FRAR. Of course, it should either use straight FRAA or maybe FRAA + 5, at most, as FRAR. So while the input is flawed, the intuition behind it is correct as it agrees with what MGL has been saying. According to MORP, here is the appropriate compensation for the following players (based on 2006 off-season $$).

2 WARP - $1.7M
3 WARP - $3.1M
4 WARP - $5.0M
5 WARP - $7.3M
6 WARP - $10.1M

Again, I don’t like WARP and the dollar figures are outdated, but the scale of growth in dollars as you move up the table seems right.

Michael I appreciate your viewpoint. I have to wonder however, if there’s a salutary effect moving from being a spot starter/ ninth inning defensive replacement , to that of being an everyday regular as was the case for McDonald this season. You’d think it would help your hitting, and likewise your fielding if you knew you’d be in there everyday. Certainly from a subjective viewpoint his defence seemed to improve when put in there day in/out and wasn’t worried about “screwing up”, which is the bane of most part time players.

I’m not sure his strong 2007 defensive season can be casually disregarded because of this factor. The truth probably lies somewhere in the middle, I can see him being a + 1 defender. Paying $1.9 million for this level of ability is not particularly outrageous, although the point is taken that teamate Ray Olmedo could well provide roughly the same performance for a lot cheaper. I suppose the old adage “ a bird in hand is worth two in the bush” sums up the teams thinking in this matter.

I tried playing around with the Morp formula, using a more realistic replacement level, modifying the dollar amounts until an average player (2.0 wins) gets paid around 8 million.  But when I do then a Pujols like player (7 wins) 64 million.

I could try to come up with a different formula, but the linear one works well enough.

Rally: Please email me this formula immediately.

Re: MAcDonald.  I actually played this “game” here at post 12:
http://www.insidethebook.com/ee/index.php/site/comments/the_greatest_fielder_ever_who_couldnt_hit_a_lick/#12

I had him at a 1 WAR player.

If you had to come up with a list of say 10 players that UZR misses the boat on, Ichiro and MacDonald are probably two of them.  I provide some simple evidence in post 15, plus the Fans assessment.  It’s just too impossible that a player this atrocious as a hitter could carve out such a long MLB career across several managers to be anything but a great glove. Either that, or he must have some fantastic clubhouse chemistry.  In any case, you have to give that some respect.

I discussed MORP in this blog somewhere.  Use the search.

What it comes down to us that because of the incredibly poor baseline that WARP uses, you need to make an “exponential” formula to gte the salary right.

In fact, for most cases, as I showed in that thread, WARP+exponentialSalaryConverter gives you exactly the same values as LWTS+linearSalaryConverter.

Here’s the thread showing how two wrongs make a right (WARP/MORP):
http://www.insidethebook.com/ee/index.php/site/comments/supervorp/

That still does not “solve” the problem of a player’s value likely not being linear with his WAR because of supply and demand as I already mentioined (and gave a good example of an extreme, hypothetical case).

I think that using an exponential model preserves reasonable salaries of players like pujols and A-Rod, but makes the salaries of players in the 1 WAR range much less than 5 mm (4.4mm plus MLB minimum).

The problem is if you fit an exponential model to do exactly that, you are going to be way low on the 2-3 win over replacement guys.

MGL: I don’t believe that your model actually exists in MLB.  Even if you want to overpay because of the supply/demand, there is still going to be a fear that you are putting so much money into one player.  Plus as the money goes up, the number of teams bidding goes down.  The end result is that you will end up with a linear link between WAR and salary. 

Since ARod is the top-end player, and is very durable, this will be tested.  If I’m right, he will be signed at 8/252 (or whatever the salary calculator says at WAR = 7).  Unless you have him at some other WAR level, of course.  I think you will find that if you take the top 5 WAR players out there, they will get paid according to a linear method.

***

Romero: signed for 3/12, paying for 1.36 WAR.

Talk about getting lucky!  He stranded 90% of his runners last year!  Santana, Peavy et al don’t do this.  I could look at this splits with men on base, but I already know what it’s going to tell me.  What the heck:
http://www.baseball-reference.com/pi/psplit.cgi?n1=romerj.01&year=2007

With men on, a .208 SLUGGING!

With bases empty: .250/.378/.446.

His career splits:
men on: .275/.375/.408
empty:  .235/.333/.364

(You have to take some care because of the IBB.)

On the other hand, he’s got great WPA.  If you look at those numbers, he’s a .600 reliever.

If you look at his FIP, he’s probably a .500 reliever. 

He pitches under 60 innings a year, which means you get 6 full games out of him, maybe 7.

With replacement level at .470, his implied WAR of +1.36 gives him an implied winning % of .670 to .700.  He’s definitely overpaid.

His WAR is 0.2 to 0.9, depending on how much faith you put in his WPA.  The most you should pay this guy is a 2/6 deal.

Mariano Rivera:

He’s the kinda pitcher where his career totals is what his current talent level is.  Very impressive for someone about to turn 38.

If you want to call him a .650 pitcher, I won’t argue.  .700?  No argument.  .750?  No argument still. 

A replacement reliever is .470.  And following Guy’s method, any win about the .570 level is multiplied by the pitcher’s Leverage Index (i.e., doubled).

Therefore, Mo’s wins are worth anywhere from +.30 wins per 9IP to +.50 wins per 9IP, above replacement.

We give him 8 full games, so he’s a +2.4 to +4.0 WAR pitcher.  For a three year deal, that’s 27 to 51. And for a 4 yr deal, that’s: 32 to 65.  I basically won’t have a problem with whatever he signs. 

If I had to bet, I’ll presume he’s a 3 WAR pitcher (implies a .700 pitcher), then look for 3/36, 4/45.

Heyman said 3/42, which implies a 3.4 WAR pitcher, or .710 (if 81 IP) or .730 (if 72 IP) reliever.

Tango pegs Posada at 4/55.  Heyman says 4/52.  The Yankees say…

...4/52.

Scary.  Or boring.  Not sure which.

I’m kinda slow, so let me see if I get this.  Todd Jones just signed a 1 year deal for 7 million implying 1.6 wins.

His FIP last year was 3.77 and the AL FIP was 4.51 making Jones a .580 pitcher.

He threw 61.1 innings so we call it 7 games.

He gets .110 wins for being above replacement as a reliever, and then a small chunk more for being a closer(.010*2.22) so his wins would be .132*7 games making him a .9 to 1 win pitcher.  And the Tigers significantly overpaid.

Billfer: thanks for doing my work!

I should caution that you don’t just look at one year’s work.  We are after true talent, so you look at the last few years, and include aging.

Note that his 3.77 FIP is actually is Bill James forecast for 2008 (which is actually a better thing to take than his 2007 FIP).

Not that it matter for him, since he gets little value out of it, but I would just give a blanket 2.0 LI for any closer.

***

Jones: signed 1/7 paying for 1.6 wins; over 7 full games, and LI=2 implies .635 win%.

Over his last 4 years, his WPA/LI has been: 3.54 over 280 innings, or +.114 per 9 IP, or .614.

(Note: I’m not positive that Fangraphs has the baseline correct, which makes a big difference.  It might overvalue all pitchers.  This has been discussed elsewhere in this blog.)

In any case, his .614 performance is from age 36 to 39, and he’ll be 40. 

I think it’s reasonable to call him a .575 pitcher, which is what billfer basically did.

He’s a 1 WAR pitcher, meaning he should get 4.8 MM.

(I just realized. I forgot to add the minimum salary to all the players, and to the salary chart.  I’ll fix that.)

Dan #65,

Welcome to Carmax style free agent pricing.  No need to haggle or negotiate.

Mike Cameron:

A slightly above average hitter.
An above average fielder
A CF.

Fielding: From 2006-2007, Dewan has him as +16 runs total.  From 2003-mid07, MGL has him as +13 runs per 150 G.  The Fans have him as a “76” (CF avg is “60"), which makes him +11 runs.  All agree: he’s +1 win.

Dude plays in a killer pitcher’s park.  He’s probably +0.5 wins as a hitter (maybe someone can help me out here).

He gets his 0.5 win bonus for being a CF, and loses it for aging.

Add it up, and we get: +1.5 WAA, or +3.5 WAR.

He’s in the same class as Posada and Lowell.

Of course, he’s got the suspension.

Tango - where do you see Bay and Freddy Sanchez?

Bay:

What happened to him in 2007?  Was he injured? 

Not only did his batting numbers plunge, but the Fans thought his fielding plunged by about 15 runs.  Dewan’s numbers also plunged by 21 plays (about 15 runs).

I don’t think it’s worth doing him, since I’d have to know how real 2007 was, and how persistent that will be.  This is one to be decided by the scouts, not the performance analysts.

***

Freddy Sanchez: 4+ MLB service time (meaning he would get 60% of the free agent value).

Average fielder, a bit above average hitter, average position, no aging to apply.  Probably +0.5 WAA and +2.5 WAR.

2/15, 3/23.  What’s the word on the street?

I had Sanchez -.5 wins fielding staying at second and +1.5 wins hitting. Toss in -0.5 coming off shoulder surgery playing up the middle and that makes him a +0.5 WAA guy like you suggested. It’s hard to sign him more than one not knowing if the injury bug is going to bite him.

1/4 in Pittsburgh with Coonelly at the helm.

I’ve pretty much done my Slwts projections for 08.  I’ve got Cameron and Bay as 3 wins above replacement (WAR) before positional adjustments and Sanchez at 2 WAR.  Sanchez gets nothing as a 3B, Bay gets deducted almost 1 win as a LF and Cameron gets .5 wins, so:

Bay 2 WAR
Cameron 3.5
Sanchez 2

1.5 win difference between LF and CF is rather large.  Didn’t you confirm my findings that the difference was roughly 10 runs?  Or are you figuring that LF was especially poor fielding-wise in 2007/08?

Mitchel - follow me a second.

Take A-Rod last three. If he averaged 97% playing time (95% and above to me is 100%) averaging 50.7 earned runs above average per year (players with 90 or more PA, MLB 2005-2007), that tells me he’s a raw 5.1 win guy in the box. Taking it one step further for this forum, if I add Tango’s 2.25 wins above replacement (97% = 2.18) that puts him at 7.4 WAR offensively. I don’t have BiP defensive metrics worth a can of corn but I’ll suggest A-Rod is +.5 defensively putting him at +7.9 WAR, and I’ll add a tick for his speed game putting him right around +8 WAR.

Jason Bay averaged 95% playing time last three and averaged +42 earned runs above average, or +4.2 WAA. Add Tango’s 2.25 (2.14) and that puts him +6.3 WAR offensively. Bay’s speed game is disfuntional anymore so no credit there, and his defense, again as a guess, was -.5 last three making him +5.8 WAR overall.

You have Bay +3 WAR and then hit him another -1 WAR as a LF.

Am I that far apart in my projections, or are you seriously discouting Bay because of 2007’s production?

First, you are ignoring regression.  Secondly, you are ignoring aging.  Finally, you are not weighting the most recent season more.  In short, you broke the only three rules Marcel adheres to!

And the run-to-win is more like 10.5 or 11.0 to 1.

Kenny Rogers is exploring his options now.  Bill James projects a 4.57 FIP in 145 innings (over 23 starts).  Essentially Rogers is an average pitcher.

Over that many innings Rogers looks to be 1.9 WAR.

If you dock him -.5 wins for being old, that puts him at 1.4 at $6.5 million.

Now what is the view here on incentives?  If Rogers actually does achieve the projections should he have a right to earn back that half win age deduction?  Is the deduction more due to the risk of not acheiving the projected talent level, or is it due to teams perceiving less value for an old player in the market place?

If Rogers stays healthy and productive for 30 starts (another 44 innings) that’s another .6 wins above replacement.  Does it make sense to have another 2.5 million kicker for achieving that?

The 0.5 wins drop is the natural progression of aging.  Basically, if you have a 2.5 WAR player between Apr 1, 2007 and Oct 1, 2007, then you should expect a 2.0 WAR player in 2008, 1.5 in 2009, 1.0 in 2010, 0.5 in 2011, and out of baseball in 2012.  That’s a general rule.  Guys at 30 age better than guys at 40.

I’m a big fan of incentives.  It makes life easy, since you are mitigating the risk.

Curous, ditto exactly what Tango said.  Obviously for A-Rod the weighting of the seasons raises his 3-year average, but the aging and regressio is a lot.  Plus his defense has been bad at 3B for 3 years now.  I have him projected at -3 runs per 150 in defense.  His baserunning “tick” is only 2 runs.

Tango, I probably have like -7 runs for LF and +3 runs for CF, so that if we round up each one, it looks like a 1.5 win diff but it is really 1 win.

Also, I keep forgetting about the arms.  For UZR, LF is better by a few runs.  With arms, it is about even between LF and RF.

Also, I agree that the poitional adjustments should not be based on average hitting at each position, as you told me many years ago, but…

If we don’t have any data on players who switch positions, we have to use hitting data to figure out the poitional adjustments.  Since the data we have on players switching positions is wrought with problems (selectve sampling, sample size, aging, etc.), we have to use the batting data at least somewhat in doing the adjustments.  At least that is my thinking.

I agree with the “somewhat”, but I’m thinking “not at all” when it comes to the three OF positions.  Those three positions all use the same skills for the most part (their ankles, feet, legs, and hands), and the arms to a varying degree.  We should know exactly how a CF would perform in LF or RF.

As for 2B/SS/3B, that is a bit more problematic, so I’m willing to accept some hitting numbers to use.  But, not much.  The talent level at 2B and 3B fielding-wise is so close, nowhere near the gap in their hitting. 

But, for IF/OF, I will have to accept alot of the hitting numbers.  If the average 2B/SS/3B is say -6 runs and the LF/CF/RF is +6 runs (no idea what it is, just guessing), but my positional adjustments say the difference should be only 5 runs, I’ll have to give a great deal of weight to the 12 run gap (or whatever it happens to be).  As you’ve pointed out, and I now accept, position switches between IF/OF are almost entirely IF to OF.  Any OF to IF switches are almost entirely done by OF who were IF in the minors (Melvin Mora and a few others excepted).  While I don’t think it’s a big deal to go from OF to IF (seeing that plenty of OF become 1B), I’ll have to concede some reality to that.

For Catchers, I have to let the offensive numbers drive almost the whole thing, because they are so different.

Tango, real quick: On SOSH you said “what should cost 800 million$ based on their intrinsic value, teams are actually paying 1.6 billion$.” I was just curious...but does this mean that MLB teams as a whole are spending around $400 million per year on FA(400*2=800)?

If you look at studes’ article, 1.6 billion$ in payroll are paid out to players with 6 or more years of service.

Gotcha.  And don’t you guys think that if we were to use average offensive values for each position (normalized to league average), we should use the median offensive value over the mean? That might help in accuracy.

The mean and median are very similar, as hard as it is to believe.

Mean and median weighted to playing time re almost exactly equal.  Mean and median by players is not.

Someone sent me an email asking for more explanation of regression.  I recommend the three links at post #1 here:
http://www.insidethebook.com/ee/index.php/site/comments/selective_sampling_how_not_to_choose_players/#1

The Mets have reportedly signed Luis Castillo to a four-year deal for $25M.  No report that any of the years are option years.  Castillo was 31 last year, but seemed older to me, probably because of his reported leg problems.  He is not longer an above-average fielder (if he ever was one) and his entire value is in his speed and bat control (groundballs, bunts).

By WSAB, he was one WAR this year, almost two WAR last year.  Tango’s chart doesn’t go four years for one WAR, but for two WAR, it shows a value of $24M for four years, close to Castillo’s contract.

But this deal is riskier than the average deal, perhaps much riskier, because Castillo’s value is almost entirely in his legs, which will probably decline quickly the next four years.  Omar’s record is a mix of smart and “huh?” deals—I put this one in the “huh?” bucket.

That is excessive.  There seems to be a good supply of over 30 second basemen who are about league average.  There was last year too.  The Cards spent about 10 million on Adam Kennedy.  While he had a bad year that deal looks like a bargain compared to Castillo.  But when the dust settled there was still Ronnie Belliard looking for a job, with a track record certainly not inferior to Kennedy’s.  He signed a one year deal for under a million.

Castillo: 4/25, paying for 2.0 WAR.

The Fans have Castillo at +1 win fielding.  UZR 03-mid07 has him at a smidge above 0.  Dewan has him like UZR.  Let’s call him even, maybe +0.5 wins fielding-wise.  2B gets no positional adjustment.  Offensively he’s league average.  Add in -0.5 for aging, and he’s league average, maybe a smidge below. 

He’s 2 WAR in full-time play, but it’s likely he won’t play that much, so 1.5 WAR would be fairer.  A 3/15 deal would have been appropriate.  It’s an ok deal.

Kennedy last year signed for 3/10, which would have implied a 1.2 WAR at the time.

The Castillo deal is defensible, but that’s pretty much the most he could have cost.

I have Castillo as exactly league average, which is around 2 WAR for a full-season (162 games).  (I give a 2B .5 win for position.) With playing time considered, I agree that he is a 1.5 WAR player.

I agree that there does seem to be a lot of average to mediocre second basemen around, leading to some real bargains being signed or that can be signed.  I have Belliard as 1.5 WAR and Kennedy at 1 WAR.

Lowell just signed a 3/37.5.  To me that is the first big overpayment of the off-season (I think).  I am not as high on Lowell as the other forecasters.  He has had a weird last 4 years offense-wise.

Here are his offesnive lwts per 150
04 21
05 -23
06 0
07 20

If he was significantly injured in 05 then I probably underrate him a little.  (Injured seasons should NOT be discarded in doing a projection as they are part of a player’s chance of playing injured in the future.)

Anyway, I have him as an average third baseman overall, so 2 WAR.  He is .5 in defense, zero in offense, and almost -.5 in running.  I think he got overpaid for 3 reasons:  One, his great 07 campaign, both in garbage stats and context-neutral saber stats.  Two, he was signed by Boston, who really wanted him.  Three, he gets all the hype, including being a great “clubhouse guy.”

"There seems to be a good supply of over 30 second basemen who are about league average.”

Isn’t this a contradiction in terms?

Lowell: 3/37, paying for 3.0 WAR.  In post 19, I had him as a 3.5 WAR, meaning 3/44.  I see it as a good signing.

***

MGL, how do you have him as 0 for offense?  The numbers you post gives you a simple average of +4.5 per 150.  If you weight on a 5/4/3/2 basis, you get +5.0 per 150.  That’s +0.5 in hitting, +0.5 in fielding, and -0.5 on baserunning (which I hadn’t considered), making him +2.5 WAR using your numbers.  Also note that because you have a higher replacement level, then the $ per win for you is higher.  I think it was 5.5 or 6.0 MM per win for you, making it a fine deal.

I think for you to make the comparisons as I’m doing, you should tell us what the average team’s wins above replacement is for you.  I’m presuming that it’s +24.3 for you (.350).

I’m using +32.4 wins (.300).

As you can see, each mglWin costs 33% more than each tangoWin.  If I’m using 4.4MM per win, you need to be using 5.87.

I agree with Tango regarding the Lowell signing.  The Red Sox are my favorite team and I follow them the most and given that Lowell is a FA (and will be overpaid no matter what, technically), signing him for 3 years (rather than 4) and less than $40 million is about as good as we were going to get.  There’s no way in hell Lowell would take $30 million or less as a FA.  Will he produce the same numbers as last year? Probably not, but I don’t see him putting up numbers similar to his last year in Florida either given that he plays half his home games at Fenway Park.  The only thing about Lowell is his defense...last year he made more errors and didn’t seem to be as “solid” as the year before (subjectively speaking), but if he can play slightly above average defense I think we’ll be fine.

Isn’t MGL’s replacement level like 17 or 18 runs below average? Does that constitute a 5.5 or 6.0 MM per win number? If so that’s definitely surprising given that just a few years ago we were at $3 mil/win.  Interesting.

You may be right.  You calculated it so that a replacement level team has a .300 winning pct (48.6 wins).  With a replacement level team playing at a .300 winning pct. and an ~ $11 million payroll, and with the average MLB payroll about $80 million, the extra $69 million will buy a team 32.4 wins ((.500-.300)*162), or ~ $2 million per win.  And then you just got the 4.4 for FA from there.  So, MGL’s may have to be higher…

Aaron Rowand:

He had a hitting season opposite of Andruw.  Bill James is forecasting a 2008 just a shade above his career average.  Including aging, let’s make him +0.5 offense.

Fielding-wise, Fans have him as an average CF (the same fans that think that Victorino is a fantastic fielder, and Bourn almost as good… imagine, having three CF, and you put the worst one there). 

UZR of course had a love affair with Rowand for a while, with him being +16/162 from 03-mid07.  However, since 2006, he’s been a bit below average.

Dewan had him as the best CF in 2005, and one of the worst in 2006, and below average in 2007.

Clearly, The Crash has had an effect on him, according to Fans, UZR, and Plus/Minus (but not his manager).

Including aging, calling him a bit below-average fielding CF is the right thing here.

That puts him right around average, plus the 0.5 for the CF positional adjustment, and he’s +2.5 WAR in full-time play.

3/30, 4/36.

I wouldn’t be surprised if a team will be signing for both his legacy as a great fielder, and his great hitting in 2007.  aka, he’ll get the Gutsy White Guy contract.  Who gets the Underachieving Black Guy contract?  Because, obviously, you can’t have the converse. 

(For your P.C. types, this was said with full sarcasm, poking a finger at the double-standard faced by Black athletes by White America.  Being White doesn’t mean you are part of White America.)

Speaking of which…

Eckstein:

Hitting: -0.5 wins
Fielding: -1.0 wins (Fans), +1.0 wins (UZR), 0 to -0.5 (Dewan).  Let’s go with -0.5.
Position: +0.5.
Aging: -0.5.

Overall: -1.0 WAA, +1.0 WAR.

Sign for 2/8.  Close to your last chance to see this guy.

Btw, moving him to 2B makes no difference.  He ends up going from -0.5 as a SS, to league average as a 2B.  And he gets no bonus for being a 2B.  So, his pos+fld = -0.5 either way.

Orlando Cabrera:
Fans think he’s a great fielder (+1.5 wins)
UZR thinks he’s a bit above average
Dewan thinks he’s a bit below average

As a hitter, he’s a bit below average.

His aging and his position cancel out.  So, you can argue for him as being anywhere from -0.5 WAA to +1.0 WAA, or +1.5 WAR to +3.0 WAR.  It really depends on what you think of his glove. 

***

Jon Garland: he’s a bit above average, say a .525 pitcher.  He eats innings, making him say 20-22 full games.  So, over the .380 repl level, that makes him a 3.0 WAR player.

***

Garland is 5 years younger, but he’s a pitcher.  In terms of future value over the next say 3 years, I don’t know what’s better: a pitcher that is 3.0 WAR now, or a nonpitcher who is 1.5 to 3.0 WAR now.

***

In order to analyze this deal, you have to know the mortgage value of each player. 

Garland is due 12MM in 2008, then he’s a free agent.  This has an implied 2.6 WAR, so he’s actually being paid just about right, maybe a bit too low.

Cabrera is due 9MM in 2008, then he’s a free agent.  This has an implied 2.0 WAR, so again, depending on how you see him, he’s right around where he should be.

***

As a 1-yr deal, this is fairly break-even.  If each team has other strengths to leverage (allows say someone else to play SS for the Angels, and opens the door for another pitcher for the Sox), then this is good.

This has got to be as fair a trade as you’ll find.  And since it’s only for 1 year, as low risk as it gets.

I moved all the replacement-level posts to a new thread here:
http://www.insidethebook.com/ee/index.php/site/comments/this_weeks_replacement_thread/

Please continue those discussions there.  This thread should focus on the player movements, and/or finances related to the moves.

Here are my thoughts on the Garland/Cabrera trade, FWIW. The best I can tell, Cabrera and Uribe are about as even both offensively and defensively as you will find. The Sox were fooled into thinking that Cabrera is a nice upgrade because his style/shape is more like a classical SS. It is strongly expected that the Sox will now trade Uribe for a couple prospects, and promote a prospect to take Garland’s place (after chaining, of course).

Assuming that Cabrera vs Uribe is a wash, and that Garland’s replacement(s) will post a 5.00 ERA, the Sox will wind up with a couple prospects and 9 million and lose 1.8 expected wins. Assuming they spend the 9 mil on FA wins at 4.4 per, they end up with a couple prospects and +.25 wins. That seems OK until you consider that if they had simply kept Uribe and traded Garland for a couple prospects, they would have saved around 11.6 mil instead of 9, and probably gotten better prospects, to boot.

But, there is another layer to consider. Uribe is not a favorite with the Chicago media because he bats .240 and is a bit paunchy. Cabrera, OTOH, is flashier and apparently more of a leader type. So, the swap should stimulate more fan interest, which should show up in the bottom line.

So, Ken Williams is essentially treading water, while boosting his own image as a motivated and active GM.

I talked about Uribe in post 32.  He’s a 1 WAR player, and costs the Sox 4.5MM.  Cabrera will give the Sox an extra win over that, maybe more, and are going to pay the extra 4.5MM for it.  The problem of course is that they are paying Uribe too.  They definitely now need to trade Uribe, or move him to 2B.  Who do the Sox have there?

The B James projections for 2008 has Cabrera at 709 OPS and Uribe at 723 OPS. Cabrera is expected to play 5 extra games, so overall it’s about a wash on offense. On defense, all I have is some of the Dewan plus/minus numbers. In 2007, neither Cabrera nor Uribe is on the 11 best list, or on the 6 worst list. For 2004-2006, Uribe is in 10th place at +20, while Cabrera is somewhere out of the top 10. And for 2003-05, both Cabrera and Uribe are dead even at +25. I don’t see any advantage for Cabrera with the glove. And, Uribe is 4 years younger.

So, where does the extra win for Cabrera come from? Remember, Cabrera is being paid twice what Uribe is being paid.

It certainly depends on whose projections you use. I would think that one should combine everyone who is good.  Trying to get a sample of all the different methodologies would be good too.  It wouldn’t help to combine 3 forecasts that all essentially use the same methodology.  I would NOT include the Bill James (whoever does them) projections in the group of “good projections.”

I would also NOT include any defensive projections other than those based on PBP metrics, like UZR, PMR, Dewan, etc., or at least the “quasi-PBP” ones, which seem to be almost as good as the PBP ones.  And when baserunning is likely to be an issue, that must be included as well.

I have Uribe as a solid 2 WAR and Cabrera as 1.5.  I have them both almost exactly the same hitting-wise, Cabrera maybe a bit higher. I also have Uribe as .5 win better than Cabrera defensively.  I have Uribe as around .75 wins in defense and around 7 runs worse than an average SS in hitting, or around dead even (average) overall.  Cabrera is only .3 wins in defense and around the same as Uribe offensively.  They would have been MUCH better off keeping Uribe, but I agree that the perception is that Cabrera is the more valuable player and the fans do like him much better than Uribe.  However, I don’t know for sure, but I don’t think that player popularity with fans has much to do team revenue, at least not for players like that (maybe for players like Bonds, Ichiro, etc.).  I don’t think that K. Williams is a good GM in terms of being able to value established player talent.  Then again, for ANY G.M. to be able to do that, he NEEDS some kind of sabermetric model which includes a sabermetric projection.  I don’t think the WS have any interest in using sabermetric models.  Without it (saber analysis), a team simply CANNOT be nearly as good as it could be with it.

---"I would NOT include the Bill James projections in the group of “good projections”.
_______________

Maybe not, MGL, but at least they are available, and available now...hint, hint.

I’d also mention that yes, in the abstract, Cabrera will not affect attendence in the same way as Bonds, Ichiro. But given the Sox situation--an unexpectedly bad year, with lowered fan interest, they need to renew some of that with CHANGE. Cabrera is change, while Uribe is the same old, same old. And the newspapers are helping by playing up Cabrera as a gold-glove SS who scored 100 runs last year…

The Mets acquired Estrada for Mota (LoDuca is a FA).  Mota makes 2.5 next year, probably about right, maybe a little on the cheap side.

Estrada is arb-eligble for one more year and then is a FA.  He’ll probably get around 5 mil, which is around right for a FA (high for arb), as he is 1 WAR.  LoDuca is also 1 WAR, although the Mets don’t like/want him for some reason.

I don’t know who is slated to catch for the Brewers, but this basically is a kiss-your-sister deal so far.

Uribe’s OPS last year was .678, the year before it was .698, and before that, it was.712.  How the heck does Bill James forecast .723?  He forecasts 132 hits, which is higher than any of his last 3 seasons by 11!  Did they go DIPS on his ass?

Over the last 3 years, total 1000 PA, here are the SS:
http://www.bb-ref.com/pi/shareit/4QGv

Cnt Player OPS+ PA From To
+----+-----------------+----+-----+----+----+
1 Hanley Ramirez 130 1408 2005 2007
2 Carlos Guillen 126 1613 2005 2007
3 Derek Jeter 126 2181 2005 2007
4 Miguel Tejada 122 1981 2005 2007
5 Michael Young 116 2172 2005 2007
6 Jimmy Rollins 106 2268 2005 2007
7 Jhonny Peralta 105 1849 2005 2007
8 Edgar Renteria 104 1908 2005 2007
9 Jose Reyes 99 2201 2005 2007
10 Khalil Greene 98 1595 2005 2007
11 Rafael Furcal 96 2067 2005 2007
12 Felipe Lopez 94 2033 2005 2007
13 J.J. Hardy 92 1204 2005 2007
14 David Eckstein 91 1749 2005 2007
15 Marco Scutaro 90 1225 2005 2007
16 Jason Bartlett 89 1194 2005 2007
17 Orlando Cabrera 89 1963 2005 2007 ***
18 Yuniesky Betancou 88 1371 2005 2007
19 Julio Lugo 88 1806 2005 2007
20 Alex Gonzalez 86 1337 2005 2007
21 Jack Wilson 84 1768 2005 2007
22 Bobby Crosby 81 1143 2005 2007
23 Omar Vizquel 80 1885 2005 2007
24 Juan Uribe 78 1598 2005 2007 ***
25 Royce Clayton 71 1291 2005 2007
26 Angel Berroa 67 1168 2005 2007
27 Adam Everett 65 1397 2005 2007
28 Cesar Izturis 62 1023 2005 2007

Every 11.5 OPS+ points = 1 win.  As far as I see it, there’s a 1 win gap between the two hitters.

I’m sure you didn’t mean to do it, but leaving out 2004 happens to leave out Uribe’s, by far, best year, and one of Cabrera’s worst, not to mention that OPS+ is crap.

Here are both player’s MGL lwts over the last 4 years:

Uribe
7, -13, -20, -18, weighted average of around -14.

Cabrera
-20, -21, -12, -5, weighted average of around -13.

And Uribe is 5 years younger.  Hmmm....

I’d say they were dead even in offense.  Tango, I think you blew this one!

VEry interesting.  Yes, OPS+ is crap because it still undervalues OBP.  And Uribe has one of, if not the, biggest gaps in OBP/SLG.  This guy’s HR/BB ratio is off the charts.

I may have blew it, which is why I should wait until I run the Marcels before doing this stuff.

Actually, OPS+ underweights OBP, which means Uribe ends up looking too good under OPS+.

I had Uribe at -1.5 wins, which is what you have.  So, that wasn’t the issue.

I had Cabrera at -0.5 wins, which is very different from what you have.  I’ll have to take a look at him, to see why I said that.

Ok, I see what I did.  I based it on his WPA/LI, and Cabrera was fantastic compared to his overall numbers.  Over the last 3 years, he was league average in WPA/LI.

What this means is that he *performed* exceptional well to the situation. 

For example, a situation hitter, when a K is costly (runner on 3B), will not K as often.  When a walk is not as valuable (two outs, no men on), won’t look for the walk, etc, etc.  I don’t know how Cabrera did it.

If we look at all player’s LWTS and their WPA/LI (i.e. game-specific LWTS), Orlando Cabrera may likely be the best in the league.  In any case, he likely *performed* the best in the league.  Whether this is persistent, I don’t know.

And remember, the WPA/LI doesn’t give the player any credit whatsoever for the leverage of the situation.  He can’t get like +.70 wins in one PA.  WPA/LI is purely a baseball measure, as context-specific as there is.

It’s on that basis that I called Cabrera a slightly below average hitter.  I will be vindicated if his WPA/LI (which IS our test by the way) will be close to 0 in 2008.

***

When I say that it IS the test, it IS!  For example, we really, really, really don’t care how many HR, or OBP, or SLG ARod will get.  What we do care is that his performance matches the context.  If you get 0 strikeouts with a guy on 3B, this is great (presuming the rest of his numbers are good).  This does help the team win.  What we are really forecasting is the player’s WPA/LI for 2008 (or his WPA if you insist).  We are *not* forecasting his LWTS or RC.

Why is that?  Because a team pays for wins and runs.  And moving a guy over is worth paying for.  It has value. 

There’s a LWTS formula for every single game state (wOBA by Game State, i.e., WPA/LI).  That’s what teams are paying for, and that’s what you have to forecast.

Forecasting WPA or WPA/LI is ridiculous unless there is some reason to think that it is a significant skill over and above lwts.  Teams are paying for a forecast that gives their team the best chance of winning, i.e., the stats that correlate best with WPA or WPA/LI.  If that happens to be prior WPA, which I doubt, then that is what we want to forecast, using past WPA data.  More likely it is a lwts forecast that correlates best with future WPA, just like DIPS or regressed ERC correlates best with future RA.

Do you want us to start forecasting clutch hitting using prior clutch hitting stats?

Now, if you are saying that we ARE forecasting WPA with context neutral stats, I have no problem with that, but there is no reason to call that a WPA forecast.  We just call it a lwts forecast, that’s all.

My lwts, which at the risk of sounding arrogant, is the ONLY lwts to use, uses different values for K outs, GB outs, infield hits, etc.  You can’t do any better than that unless, as I said, WPA over and above lwts (basically clutch hitting) has significant predictive value, which we are pretty sure it doesn’t.

The ONLY thing that a context neutral lwts does NOT do for forecasting actual, real value for a team considering a player, is his likely slot in the batting order and the players around him (which change the lwt values of the various events).  That can easily be handled though if one wants to get so granular with a value projection for a team.

As far as OPS+ it is NOT just the undervaluing of OBP.  It is everything, not to mention that, as I said, my lwts has different values for K outs, infield hits, etc.  It is just not a great stat.

The “mistake” that you made had nothing to do with your WPA/LI.  At least I was not referring to that.  I don’t know why you even mentioned that.  You gave both players’ last 3 years OPS+ and concluded that Cabrera was a significantly better hitter from that data. As I said, by leaving out 2004, you happened to leave out a season in which Uribe hit 2.5 wins better than Cabrera.  Surely that is going to have a significant impact on their projections!

I think we are on the same page when you say:

“Teams are paying for a forecast that gives their team the best chance of winning, i.e., the stats that correlate best with WPA or WPA/LI. “

Yes, that’s exactly what we want.  Our “dependent variable”, our “y”, is WPA or WPA/LI.

How you do that is open to debate.  Prior year’s LWTS, prior year’s WPA/LI, combination of both, or whatnot, is what needs to be tested.

This is similar to correlating with next year’s ERA.  Even though his next year’s ERA is influenced by fielders, park, and whatnot, that is still the test.

***

Now, as for your version of context-neutral LWTS.  Note that you are including GB and FB, and giving them different weights.  You are doing that because we know, for a fact, that hitters have a tendency to hit GB or FB, and that each of those types of hits and outs have a different value (based on the 24 base/out states).

However, and we must agree, that every single batter and pitcher changes his approach based on the game state.  Surely they know how important getting a K is with a man on 3B and less than 2 outs.  The LWTS run values here are off the charts different.  This calls for the batter to change his approach to reduce his K to some degree, and the pitcher to change his approach to increase his K to some degree.  It’s all part of the batter/pitcher matchup.

(I know we agree that the matchup changes by *count*.  Therefore, it’s not a stretch to think it changes by game state.)

The question always deal with the persistence in the data: can we find out who changes their approach and to what extent.  How reliable is the data we have.  It would be almost a certainty that it’s easier to find this with pitchers than with hitters, since he controls the pace.  Glavine is the typical pitcher brought up.

***

The three year total of Cabrera in WPA/LI is 0.0, and mgl’s lwts is -38 runs (-13 per year).  The 4th year really didn’t matter here.  This is an enormous difference, that after some 2000 PA that we can come up with a 38 run difference.

Note that Cabrera’s LWTS by the 24 base/out states is +4 runs over those three years.  So, it’s not even that he tailored it to the game state, but simply to the base/out state.  That is an enormous difference.

MGL, how much weight would you give to a 4th year? Marcel doesn’t use it at all.  For CHONE I think it would account for something like 5-10% of his projection.  Most players aren’t changed too much if you use it or not.

I’m sure Barry Bonds would be an exception, since his year x-4 is the best season anyone’s ever had and year x-3 he barely played.  Though the biggest challenge in projecting Bonds is to estimate the level of competition he’ll face in the California Penal League.

The last two years, Uribe’s GPA has been around .220; Cabrera’s has been around .260.  I know GPA is only a little better than OPS, but I’m having a hard time swallowing the idea that they’re even offensively.

Rally:

T: 1.00
T-1: .80^1
T-2: .80^2
T-3: .80^3
T-4: .80^4
...
T-n: .80^n

For pitchers, I use .70 instead of .80.  I’m not sure what I’d use for fielders.

I’ve posted a “daily” version as well on this blog elsewhere.

Assuming 600 PA each and every year and using that framework, here’s how much a 2008 projection depends on each year of input data:

2007 30%
2006 24%
2005 19%
2004 15%
regression: 12%

The 600 PA would be:
600*(1+0.8+0.64+0.512)=1771

regression would be roughly 200/(200+1771)=10%

Unless your 200 is really 240, in which case you are right.

Tango, MGL—Ian Snell in Pittsburgh and Ian Snell in St. Petersburg (Rays).

#116 ??

Studes, I gave each player’s MGL lwts for 2004-2007.  It is park and opponent adjusted.  There is nothing more accurate from a context-neutral persepctive.  Nothing.  GPA, EQA, OPS, OPS+, etc. is not going to add ANYTHING to MGL lwts.  That’s just the way it is.  None of those stats treats the components correctly.  In most cases, they will all be about the same, but in some cases they will be a lot different.  Bu once you know MGL lwts, knowing any of the other stats is not going to add anything.  Of course you have to take my word for it.  Or you can do the work yourself.  I don’t have my “work” computer with me (I am out of town), but basically I take a player’s raw stats, I adjust each component for EVERY park the player has played in (THAT I am not claiming is perfect by any means), and I adjust for the opponent pitchers.  I don’t adjust for opponent pitcher handedness.  Then I assign a (context-neutral) lwts value to infield singles, outfield singles, d, t, hr, NIBB+hp, K outs, GB outs, with a different lwt value for RHB and LHB, and FB outs.  I don’t include ANY IBB or sac bunt attempts.  You can’t do any better than that, except perhaps include IBB in some way (that has been debated before).

I’d love to see what MGL and Tango think of the Hunter signing.  A big question is his defense—his raw PBP stats are kinda pedestrian, so was he by PMR.  Dunno about UZR.

I’d guess his offensive projection is, what, around +6 or so?  Give him around +5 or so for the position, and even if he’s an average defensive CF at this point, he’s still about a win above average, and three above replacement.  So $16M per doesn’t seem too far off, if wins are really getting up to the $5M range.

Of course, at his age he’s an injury and decline risk.  I know MGL always says that salary inflation tends to balance out aging, but when you’re talking about a CF who is 32 years old and isn’t so much on the OBP, I don’t know ... esp. when this means putting Gary Matthews Jr on a corner.  He’ll probably gain back the runs defensively he loses by position, but you’re still putting a guy on a corner who is essentially a league-average hitter.  I guess Hunter being an above-average CF with the bat makes up for that a bit.

Still, it seems pretty weird, given that you have Kendry Morales and Juan Rivera around to DH, and can give Reggie Willits a start or two per week as well, in the field, so that you can “rest” Garret Anderson and Vlad there.

Basically I’m shocked and confused and am trying to make sense of it ...

Torii: 5/90, paying for 4.4 WAR

My thoughts on Torii are at post 33:
http://www.insidethebook.com/ee/index.php/site/comments/sabermetric_moves_of_the_2008_pre_season/#33

I had him at 4.0 WAR, meaning they should have signed him 6/90 or 5/80.

And yes, you are unleveraging a couple of runs by putting a “real” CF in the corner.

Thanks, MGL.  I appreciate all the work you put into your projections.  Bottom line, however, there is no way I feel that Uribe and Cabrera are equivalent offensive players at this time in their careers.  Every reasonable and accessible batting stat would say there is a difference.

I don’t know how much weight you put on each year—that may be an issue.  The last two years, Uribe’s OBP has been .257 and .284!  Perhaps the aging curve you use is a factor.  Dunno.  But it doesn’t pass the smell or eyes test to me.

And the Hunter deal is a head scratcher to me, too.  Overpaying by a couple million per year by a team that didn’t need to.  Weird.

MGL’s forecast for Uribe in post 105 is fine.  Being 20 and 18 runs below average is pretty horrible, and his 14 runs average is consistent with what I have.  (Whitesox also play in a somewhat hitter’s park.)

The question is Cabrera:
“Cabrera
-20, -21, -12, -5, weighted average of around -13. “

***

In 2007, he had an OBP of .345, with a SLG of .397.  That’s just a shade below league average.  Angels play at a slight pitcher’s park.

He’s probably a good baserunner.  He’s a great basestealer (165 SB, 39 CS, 81%; 68SB, 9 CS in last 3 years).

So, in 2007, MGL has him as -5, which sounds a tiny bit low.  His LWTS by the 24 base/out states (not park adjusted) was +15 runs.

If you look at his splits with men on and bases empty, there is an enormous gap.  The guy was a star with runners on base with men on last year.  Now, if you were giving out “performance bonuses”, after-the-fact, you’d give Cabrera lots of money.  But, paying for the future, you can’t expect most of that to repeat (especially when you look at his performance with men on in the year or two before that).

So, no big issue with MGL’s -5 here.

***

In 2006 though, his performance was very close to his 2007, context-neutral.  I don’t see how that could possibly be a -12 performance, if 2007 was -5.

His LWTS by the 24/baseout states was 0.  His WPA/LI was a plus 0.7 wins.

I have a big problem with the -12 here.

***

In 2005, his WPA/LI was -1.3 wins, his LWTS by the 24 base/out states was -11 runs.  MGL has him at -21.

***

It’s almost like MGL looked at my numbers, and said “subtract 10 runs every year”!

MGL, can you walk through Cabrera’s 2006 performance, according to SuperLWTS.

I should probably go here more often:
http://www.baseballprospectus.com/dt/cabreor01.php

Cabrera’s Batting Runs Above Average (BRAA), which is park adjusted, are:
2007: 0
2006: -1
2005: -11
2004: -16

So, the 2005-2007 numbers are entirely consistent with what I’ve been saying.

MGL?

The Angels made news again today signing Torii Hunter for 5 yrs. and $90M. Tangotiger had him as a 4 WAR player in post 33. CHONE has him as a .361 wOBA hitter, which is about 10.7 RAA per 560 PA. So, he’s almost exactly 1 WAR as a hitter which is what Tangotiger assumed.

Now, Dewan had him as a +29 (plays) defender from 2004-6, but he does not appear on the list of best players from 2005-7. The cutoff there is +21, so let’s assume he was a -8 for 2007. PMR has him as even for 2007 and 2006. UZR has him as a +11 (runs) from 04-06 and had him as +5 at the break. Using a weighted average, the systems have him at +2, 0, and +9 runs, respectively. The fans had him as a +8 (runs) defender this year. Doing an average of the four and adding a bit of regression gets me to +3 or +4 as a defender. This is where Tango and I slightly diverge as he had him as a +10 defender.

CF is worth roughly .5 wins which is offset by .5 wins for aging. So, that makes him a 1.35 WAA player or a 3.35 WAR. That comes out to 5/60.5 on Tango’s salary chart, so this is a definite overpay.

Don’t know why you said the Fans had him as +8 runs.  He has a rating of 77, which is about +17 above CF average.  Multiply by 0.7, and you get +12 runs.  That’s why I said +1.

Now that this has sort of settled in, the signing doesn’t seem too bothersome.  I think Hunter is more likely +.5 wins offensively, +.5 for the position, and neutral for the fielding—at worst.  So that comes out to around 3 WAR as his floor.

I’m worried about years 4 and 5 of the deal, just as I am with Gary Matthews Jr, but moving from Cabrera to Hunter is an upgrade, ceteris peribus, and as it’s not my money I guess I shouldn’t really worry about the extra $7M.

I don’t have my work computer with me so I can’t walk through Cabrera’s lwts.  I am out of town and won’t be back until next Friday.  Sorry.  I firmly stand by my lwts values for both players.

Hunter is a bad signing, but then again, the Angels have a history of that and I don’t think that they have any sabermetric orientation in the front office.

I have Hunter as 3 wins above replacement.  .5 win in defense (7,6,5,10 last 4 years), 0 in offense, neutral in baserunning, and .5 in position.  He is overrated in the “conventional world” in both offense and defense.

My “decline with age equal to inflation” (roughly) is basically included in Tango’s scheme, although he has the decline as more than inflation, which is fine.

BTW, I have Matthews, who is an underrated player, as almost the same as Hunter, maybe .5 wins less, for only 10 mil a year.  Given that, the Angels really blew it.  But I am sure that almost everyone and their brother think that Hunter is twice the player Matthews is and that this was a wonderful (although a little pricey) signing for the Angels.

Now, it might be a good signing if they could not have acquired those 3 wins anywhere else for cheaper (as in Matthews and someone else).  As I always say, you can’t evaluate a signing unless you know the alternatives and/or you know how much revenue a win provides to a team.  If the Yankees get an additional win for 10 mil, they probably make money on that.  If they could not have gotten that same win, for whatever reason, for less than 10 mil, then it was a good signing despite being more than twice the going rate and a seemingly horrible acquisiton.

I have Hunter as 3 wins above replacement.

Well, if each win is $5M, he’s only being “overpaid” by $1M, which hardly seems outrageous.

0 in offense

This intrigues me:  by traditional LW, he hasn’t had a below-average year with the bat since 2003; Davenport agrees.  The BRAA at Fangraphs has him averaging +5 per 600 PA over teh past four years (not sure how they determine that, it’s not in their glossary). 

So BB-Ref has him at +9 per 600 over the last 4 years, Davenport +8, Fangraphs +5.  Though it’s not as extreme as with Cabrera, S-Lwts seems to be seeing something no other system detects.

Whoops, he’s getting $18M per, so the “overpay” is $3M, a bit more substantial ...

BRAA at Fangraphs is Linear Weights by the 24 base/out states.

WPA/LI is Linear Weights (in Wins) by the Game State (inning, score, base, outs), with the Leverage depressed.

MGL - breakdown Snell for me. Sorry about #116.. the brews, the food, the company.

Got to admit, MGL, that I’m scratching my head at your player value assessments.  Matthews only slightly below Hunter?  I agree that this contract overvalues Hunter, but to me he’s more than .5 wins better than Matthews.

In fielding, Dewan’s plus/minus system has him at +14 over the past three years, while Matthews is -5.  Not a huge difference over three years, but some difference.  Let’s say that’s six plays a year at .7 runs a play = 4 runs a year or almost half a win in fielding.

And I don’t see how their offensive value can be that close, given that Matthews’ one very good offensive year was in Texas and driven by an unsupportable BABIP.

Matthews walks more than Hunter, but Hunter has more power and hits for a better average.  I would have guessed at least a win difference in offense.

My guess is that Hunter is +5 runs offensively and Matthews +0, and Hunter +5 runs defensively and Matthews +0.  Obviously, they both play the same position.

Does Hunter get knocked because, as an RHB, his groundouts are less valuable?  Does this afflict all RHB in in S-Lwts, and particularly groundball hitters?  How does S-Lwts see Jeter’s hitting vis-a-vis other advanced metrics?

Well, UZR is UZR.  I do the same thing with UZR as I do with offensive lwts.  I do a basic Marcel.  I weight last 4 seasons, regress and age adjust.  I now regress toward a mean which reflects a player’s age and speed score (I used to not use speed score).  Obviously UZR is not the same as Dewan’s plus/minus.  What is the point of saying, “I can’t believe your defenisve rating/projections, Dewan has X?” Dewan is Dewan and UZR is UZR.  They both use the same basic methodology, but different data bases.  I park adjust and he doesn’t. They are what they are.  Other than park adjustments, combining both is probably a good idea.  And combining both with Fan rating (with Fan rating having decreasing importance as the sample size of the data increases) is probably best of all.

Yes, RHB’s get knocked on ground ball outs.  The most significant difference you will get between my lwts and a traditional one (and mine are better) are high K players will do worse, RHB with a high ground ball rate will do worse, and players with lots of infield hits will do a lot worse.

And any system that treats an IBB as a regular bb or hp, if the player has a lot of IBB, is basically worthless!  Watch out for that if a player has lots of IBB!

You would be surprised how much aging and regression affect a projection (correctly so).  If a player is past 32 years old or so, such that all of his prior 4 years are past his prime, if he averages well over zero in lwts, and his regression mean is below zero, as it would be for most CF, C, and SS, between the aging and the regression it is likely that his projection is quite a bit below zero.  So please stop saying, “How can so-and-so who plays a premium defensive position (and therefore has a below zero mean to regress towards) AND is in his early to mid-30’s (such that his projection will be less than ALL of his prior 4 years), never have a year below zero but still have a projection of zero or less!” I just told you how that can be!

Hunter’s last 4 years’ (MGL) lwts per 630 PA:

04 -2
05 +3
06 +9
07 +10

With aging and regression, that is a projection of +1.

Matthews is:

+9
0
+18
-2

That is about the same offensive projection, actually 3 runs less at -2.

I’m pretty sure that Marcel comes with about the same projection, using those numbers.

#129, we are up to 5 mil per win now?  I thought it was 4.5?  And that is for one year only, in essence.  IOW, according to my 3 WAR, Hunter should get 1/14, and I don’t know for 5 (I’d have to check tango’s chart).  But a heck of a lot less than 5/90.  I don’t know if you want to call that outrageous. I don’t think I used that word. I said “bad.” I don’t know why you implied I used the word outrageous.  But, you can call it whatever you want.  It is what it is.  According to the scheme that is prevalent in this thread, it is an “overpayment” of a lot of millions.

Snell is a good pitcher.  Almost half a win better than an average starter.  That is gold for a second year player (you get a player worth 10 mil on the FA market for .4 mil a year).  He is still fairly young, but it would be nice if he were younger (and was therefore expected to get that much better).  Why do you ask about him?

MGL, I had thought that you and Tango had been discussing wins being $5M, I didn’t recall the $4.5M.

Are we really going to play semantics on “outrageous” vs. “bad”?  Did I imply you said “outrageous”?

If a player is past 32 years old or so, such that all of his prior 4 years are past his prime, if he averages well over zero in lwts, and his regression mean is below zero, as it would be for most CF, C, and SS, between the aging and the regression it is likely that his projection is quite a bit below zero.  So please stop saying, “How can so-and-so who plays a premium defensive position (and therefore has a below zero mean to regress towards) AND is in his early to mid-30’s (such that his projection will be less than ALL of his prior 4 years), never have a year below zero but still have a projection of zero or less!” I just told you how that can be!

Okay, I will never ask this question a second time.  Just realize that from time to time S-Lwts pegs players differently from publicly available evaluation systems, so your projections are sometimes surprising, and people may wonder how you’ve come to them.

At any rate, I’m starting to feel less good about the signing again ...

Honestly, those are just my projections.  I am not married to them.  In fact, Pecota probably does a better job than I do on the offensive side.  And I am not married to UZR either.  So while I have him as 3 WAR, he may be 3.5.  He may be 2.5.  He may even be 4.0.  And as I say, just because someone appears to pay more than a player is worth based on his estimated WAR does not necessarily make it a bad signing.  Hunter is a good player.  I would not expect the Angels to know what he is worth or even think in those terms (that players have a certain WAR value and that they should be paid x amount of dollars per win).  They are not a sabermetric team.  They would not even know what a win above replacement was. How much a team pays for a player or what kinds of trades they make tells you a lot about how smart they are but it does not necessarily tell you whether they were “good” signings or trades from a fan’s perspective.  Any signing or trade that improves a team and does not preclude a better trade or acquisition is a “good” one no matter how much it cost.  Fans don’t and shouldn’t care how much a team spends only whether they improve their team or not and by how much compared to alternatives.  In fact, the more money the better.  Of course, if a team is not smart AND they don’t have a perennially huge payroll, it is not too likely that they will be consistently good.  So as a fan, if your team is not smart, they are a lot less likely to be as successful as they could be given their payroll.

The salary chart makes it quite clear I’m using 4.4MM per win:
http://www.tangotiger.net/salary.html

There’s a dispute as to how much MGL should be using.  I figure he should be using somewhere around 5.5 (this is based on repl-level pitcher as starter being 0.75 below avg starters, and repl-level pitcher as reliever being something like 0.30 below avg relievers).

A tangoWin does not equal an mglWin.  Therefore, we can’t tell how much we should pay for an mglWin.

The reason I said Hunter was +8 defensively is that the actual average for CF is 63.4, not 60 and that he played “only” 90% of a season in CF. Anyone can view the full list of CF on this spreadsheet:
http://spreadsheets.google.com/pub?key=pGUUiOCPQyLZFN61k0YriPQ

So, it’s (77-63.4) *.7 *.9 ~= 8.5.

Good job weighting the values by the IP played.

When I say someone is +1 or +2 or whatnot, it’s based on per 162 GP.

At the end, I convert down to GP.

So, someone can be +1 off per 162, +1 def per 162, and +0.5 pos per 162, and -0.5 aging per 162.  The total is +2 WAA, or +4.25 WAR.

Then, I apply expected games divided by 162 and multiply by 4.25.  So, 146 games is 90%, and 90% of 4.25 is +3.8 WAR.

Yeah, but my replacement level is 2 wins below average per 162 games.  Isn’t that the same as yours (Tango)?  Since avreage is not debatable, if my replacement is the same as your replacement with respect to average, then we should have the same $ value per win, no?

#143
Shouldn’t average replacement level be the same across all positions? If that’s so, then your replacement team wins only 31 games. If that’s not so, and only for hitters, then a team with average pitchers and replacement hitters would win about 53 games. That’s too low for a team made up in such a way.

MGL’s replacement level for nonpitchers is 16.8 wins below average, as detailed somewhere on this blog recently (19.44/10*8.65).

My replacement level for nonpitchers is 19.5 wins below average (2.25*8.65).

Presuming that the pitcher repl level follows the same proportion, each mglWin costs 16% more than each tangoWin.  My free agent win is 4.4MM, making each mglWin 5.1MM.

***

Dan/144: MGL’s team of all-replacement players (nonpitchers, pitchers) would win something like .325, .330.  Mine is set at .300.

I say again, my replacement level for non-pitchers is by definition 20 runs per 162 games below average.  I don’t know that means team-wise or win-wise. I don’t think in those terms.

I don’t know that that is what replacement level is.  I just use that because it is probably reasonable and it is simple and also the worst projections I seem to have for the bad position players who still hang around the majors is around that.

I’ve asked this before and I either never got an answer or I don’t remember the answer, but where do you (Tango) get your replacement level from?  And how many runs below avreage per 162 games (700 PA) is your non-pitcher replacement level?  Every time you give us .300 or .320 you confuse us (at least me).  How about you just give us runs below average (per whatever) for non-pitchers and runs per 9 below average for pitchers (starters and relievers separately if you like).

I know we have a replacement thread, but just give me your short answer.

I was wondering if you could determine the value of Cordero for the next 4 years to see whether it was a good signing for the Reds. Would you think he’s a 3 WAR pitcher and deserves a 4/45, making it a reasonable signing?

Mark, I think what you’re wanting is present value.

Tango, from what MGL is saying (and has been saying), his replacement level is either 18 runs or 20 runs below average, which is basically what you’ve been saying (right?).  So, I don’t think there’s that 16% difference that you believe there to be...I understand you’re reasoning if MGL’s numbers were way different than yours (because a $5.1 mil/win is a lot different in terms of money/salary than $4.4), but I don’t think you guys are that far off…

Mark 147: I’m not very good at this sort of thing, so I’ll wing it. His ‘06 WARP from BP is 6.0; a quick off-the-cuff estimate based on MORP is that 6 WARP is worth between 2 and 2.5 TangoWAR, so on a four-year deal he should be making $24.3 million.

An extremely off-the-cuff weighted average of his last three year’s WARP puts it at about 4.5 WARP, or (very roughly) a 1.5 WAR reliever, so worth no more than a 3-year, $14.1 million deal.

These are all real off-the-cuff numbers, and worth a grain of salt. The larger point here is that it’s extremely hard to get a gauge on the true talent levels of middle relievers—the reason so mant of them seem to flame out so quickly is that is you only pitch 80 innings a year it’s a lot easier to luck yourself into a good year than it is if you pitch 200+. This makes any forecast of them much riskier than projecting starters (which is still risky) or hitters. Long-term deals for relievers, if you ask me, are unnecessary risks even if the salary-to-wins ratio looks right using the best projections available.

Well with the Reds standpoint, adding a pitcher like Cordero replaces a RP that is basically below the average replacement. It’s not as if they are replacing him with a lost reliever; they would be getting rid of the worst reliever spot.
I know relievers are harder to judge since there’s so much luck involved due to lack of IP. That’s why I asked in the first place, and kind of used the Rivera post to guess he would also be a 3 WAR pitcher.

Mark,

I just made a post on my blog that tries to ballpark what his salary “should be” using Tango’s methods.  I’m sort of new to thinking about contract values in a rigorous way, but my most optimistic estimate is that Cordero has averaged about 1.4 WAR over the past three seasons.  That would put him at just $6.5 million on a one-year deal, which is a heck of a lot less than the $11.5 million/season he’s getting over four years with the Reds… And that’s not even accounting for the age-related dropoff you have to expect from him.

Here’s my post on Cordero, if you’re interested:
http://jinaz-reds.blogspot.com/2007/11/reds-sign-francisco-cordero-probably.html

As a Reds fan, I’m excited to have the help in the bullpen… but yikes.  I hope my numbers are massively screwed up somehow, but I unfortunately think they’re probably pretty close.
-j

Cordero, Francisco that is, is a stud reliever/closer.  One of the 3 or 4 best in baseball right now.

He is around .5 runs per 9 better than the average closer, which is worth around an extra .8 wins or so (above an average closer).  I don’t know what an average closer gets paid on the FA market.  Maybe 6 or 7 mil?  So Cordero would be worth 8 to 9, which implies he is worth 2 WAR.

At this point I don’t know what a replacement reliever is as compared to average.  I really don’t.  I would think that you could take a replacement pitcher, turn him into a closer and he would be a run worse than an average closer at worst.  Maybe .75 runs.  At a run worse, that would mean that a replacement closer would be 1.7 wins worse than an average closer, so that Cordreo would be 2.5 WAR.

So let’s say 2 to 2.5 WAR, whatever that is worth for 3 years.

I don’t know, but if I am not even close to being a contending team, which the Reds are not, I’m not sure that I would be spending a lot of money on a closer.  The very first thing the Reds should do is to try and trade Griffey for ANYTHING they can get, literally.  He has ZERO value as a player.  He is worse than replacement level.  IOW, if you could trade him to another team, get NOTHING in return and subsidize 90% of his salary, that would be a net gain for the Reds.  Surely they can do better than that.

MGL/146: you’ve told us your per-player (nonpitcher) is 19.44 runs per 162.  I’ll presume that’s 1.944 wins.  Mine is 2.25 wins.

2.25/1.944 = 116%

Griffey is only below replacement level when he’s getting a huge negative value from his fielding.  There’s a floor to how much negative defense/position value a player has, and that is -15, the value of a fulltime DH. 

Griffey should be DHing and has enough bat to help somebody at that spot, but if the Reds want to trade him they have to convince Griffey to waive his no-trade rights as a 5/10 man.

Then again the Reds would probably have to eat some salary even if a trade was agreed to by all parties.  I’ve got Griffey around 1 win above average hitting, and a replacement level DH is an average hitter.  Maybe its -5 (I just said -15 for DH and replacement is -20) So he’s +1 to +1.5 wins above replacement at DH.

He’s worth 5-7 million or so but makes 12-13.

Hmm, I had Griffey at about 26 runs above replacement this past season (hitting + fielding + position adjustment).
http://jinaz-reds.blogspot.com/2007/11/player-value-part-4-position-player.html
Moving to right field helped him a great deal this year, as did the return of his on base percentage. 

Granted, my fielding estimates are a bit bastardized (I used a weighted mean of RZR, ZR, and Fans to get that number--I’m planning to substitute PMR for RZR, but that will just help Griffey).  But I’d be surprised if they’re off by 25-some runs.  He may decline a bit next season, but I’d be surprised if he goes below replacement-level. -j

Rally, why do you say there is a floor to defense futility and it is -15?  Sure, we generally assume -15 to +15 and that may be 3 SD in talent or so, but I hardly think there is a floor and SOMEONE is going to eventually be outside of 3 SD (.3%, right?).

If you put D. Ortiz in RF, you don’t think he is going to be -30 or -40 runs in defense?  Here are Griffey’s UZR the last 4 years, some in CF and some in RF:

-26 CF
-41 CF
-34 CF
-13 RF

Convert those CF numbers to RF (add +8 or +10 maybe?), take a weighted average, age adjust, and then regress to a 39 yo player with a terrible speed score, and I guarantee you are going to get a number a low worse than -15.

His arm is horrendous.  Here are his arm scores for the last 4 years:

0
-3
-2
-10

So even if we give him -15 in UZR in RF, we have to add -5 for his arm, which is -20 in defense. 

His baserunning last 4 years is:

-3
-4
-3
-7

That is a projection of -4.

So we now have -24 in baserunning and defense.

It’s not looking too good for him being above replacement.  An average RF is a win above average in hitting.

Here are Griffey’s last 4 years offensive lwts:

5
31
-4
12

That is a projection of +4, which is 1/2 win LESS than an average RF’er.

So he is 3 wins BELOW an average RF’er, which is around 1 win BELOW REPLACEMENT, by far the worst full-time player in baseball and possibly one of the worst full-time players of all time.

ONE WIN BELOW REPLACEMENT.

As a DH, you get rid of 2 of those wins.  If a DH replacement player is the same offensive level of a replacement RF’er, then Griffey is 1 WAR as a DH.  But the fact remains that CIN is playing him as an outfielder, which is one of the most egregious and costly mistakes a team has ever made with regard to one player (paying a player 12 mil for performance under replacement level).

Rally, saying that a replacement DH is an average hitter is kind of odd when an average DH is less than 1 win above average.  You are saying that a replacement DH is less than a win less than average?  That may be, but it would be odd.

MGL, thanks for the breakdown.  I don’t include baserunning (aside from SB/CS) or throwing arm in my estimates, so that’s a big part of our difference.  Using your numbers pulls him down to 9 RAR in ‘07 using my fielding and offensive numbers.  The throwing arm is particularly surprising to me, as Fans seem to rate his arm as a touch above average (could be the veteran player effect, of course).

To account for the rest of the 2 wins difference in our estimates, I have Griff at +16 RAA on offense (not sure why we’d differ, but it’s only 4 runs), and -7 FRAA in ‘07 (Fans and ZR put him at about average, RZR has him well below average), which combined is about a win better than you have him in ‘07.  And your offensive projection is 8 runs below your estimate of his ‘07 production, which accounts for the rest of the difference. 

So anyway, that explains why you and I had such a big discrepancy.  Probably didn’t bother you much, but it bothered me!
-j

I’m curious as to what MGL is using for his arm ratings. I know John Walsh did some work for the THT Annual on throwing, so that may be it. Anyone?

Phil, I do all my own ratings, including arms.  It is the “reverse” of baserunnng and is holds and assists as compared to an average fielder at that position, park adjusted.  Griffey’s poor arm rating is probably as much a result of an arm that is not nearly as good as it used to be as he takes a lot of time getting to a ball that is a hit.

Jinaz, there is no chance (well, maybe .1%) in this universe that Griffey is a -7 in fielding.

People sem to forget about the strong effects of aging in a player’s late 30’s.  Lete’s say Griffey was a true +10 over the last 4 years.  Just with aging alone, he would be a zero next year.  A player in his late 30’s loses around 4 runs a year in offense (I think, OTTOMH).

It seems to me the floor for defensive futility is largely whatever the manager is willing to tolerate....

I must not have explained myself very well and my concept of a defensive floor.

I am not disputing that Griffey is probably worse than -15 in either right or left, that David Ortiz would be -40 in the outfield, or that Frank Thomas could be -75 playing shortstop.

But lets say the position adjustment for DH is -15.  That’s what I use anyway.  Any good hitter/terrible fielder could DH somewhere.  Griffey’s best value would be on some team that would DH him. 

Say you’ve got a player who’s +30 on offense and -50 on defense.  You can’t just call him an overall replacement level player who has no value.  Maybe he has no value to the team that keeps sticking him out in the field, but as a free agent or trade target, he damn well has value and quite a bit of it, somebody could use him as a DH.

I guess if I use -15 as a positional adjustment for DH then then a replacement level one would be -5. 

But its almost certain that the gap between average and replacement level and average should be less at DH than at other positions.  Your pool of available players is bigger, it includes everybody.  and you don’t always have truly outstanding hitters there, as usually the very best hitters are great players in their prime, and probably capable of playing a position. 

At least that’s the way I think its been in the past, right now we’ve got two DH’s who rank among the very best hitters in the game -Hafner and Ortiz.

Jinaz, there is no chance (well, maybe .1%) in this universe that Griffey is a -7 in fielding.

I dunno, I’m not sure that he isn’t a -7 fielder (pre-position adjustment), at least in terms of tracking down fly balls.  Fans have him as ~average overall in RF, and PMR has him at ~-10 runs.  That said, I’m not sure that he isn’t a -15 fielder either.

Your comment about reduced “arm” ratings due to being slow to get to the ball makes a lot of sense, though.  Unfortunately, that’s not something that I can track at this point, at least not yet.

Anyway, I agree that the Reds would be better off trading him if they possibly could.
-j

I don’t disagree that the pool of players available for DH may be quite large which means that there may be a small gap between average and replacement.  Not to mention the fact that many teams do not even use a “pure” DH which means that the average DH does not hit nearly as well as they could/should.

I see you what you mean now by the -15.  Sure, a player that is terrible defensively and is a good hitter can have MUCH more value as a DH.  Manny is another example of a player who would have MUCH more value as a DH.  Normally when you move a player to the left in the defensive player, you merely add 5 to 10 runs to his defensive value (not including the positional adjustment of course).  But in the case of the DH, it doesn’t matter how bad your defense is.  You lose all your negative defensive “value” and the positional adustment is only -15.  So, as Rally says, the most you can “lose” from your offensive value when you are a DH is 15 runs (actually you ALWWAYS lose exactly 15 runs).  For any other position, you can lose as many runs as your defense is bad.  So I see what you mean by a “floor” of 15 runs as a DH.

But the fact of the matter is that Griffey plays the OF and is on an NL team so until if and when he goes to an AL team as a DH, he is a complete disaster.

I think we’re all in agreement he should DH.  A return to Seattle might be possible, though they have Raul Ibanez, another guy whose best position is DH, and probably think Jose Vidro is a good option there for some reason.

The Twins got nothing out of DH last year, he’d be a very good fit there, but as a 5/10 man, Griffey would have to agree to any move.  He probably is not of the opinion that he should be DHing (Its quite possible he still thinks he can play center).  The athletes themselves are always the last to realize their limitations.

Yes, I’m with Rally on this.

The floor should be -20 runs.  The position adjustment for a 1B is -10 runs, and a bad fielding 1B that you can tolerate is -10 runs relative to the average 1B.  Any worse fielder, and he’s a DH.

The *only* reason we are doing an extra +5 runs for the DH (-15, instead of -20), is because of the “DH penalty”, similar to the “PH penalty” that we found in The Book.  (The opposite of the Reliever premium.) It’s harder to DH than not, and therefore, you need to compare Hafner’s and ORtiz’s replacement based on how that replacement would do in that role (which means your performance stats take a hit).

So, the bottom position+fielding that I take is -20 runs relative to average.  And, I make a separate adjustment of +5 runs for the DH.

Does anyone remember when Griffey was taken out of a game in the 9th for a defensive replacement and he (purportedly - nothing you read in the media is EVER necessarily true) told his manager, “Don’t ever do that to me again?”

I lost all respect for him after reading that, assuming that it is true.  One, for being clueless about his own defensive ability.  If you are THAT bad and don’t know it, you are delusional.  Two, for saying that, regardless of what you think of your ability.

I give him a pass on the delusional stuff but not the prima donna crap.

Steve Finley still thinks he can help a ballclub at age 43, despite not being able to do so since 2004.  Probably at least half of major league players would never have made it to where they are if they had listened to people telling them they didn’t have what it takes at some point on the way.

The player is going to think he can play until everybody tells him to go home.  Its up to the managers and GM’s to make a rational decision on when to tell him that.

I don’t remember the incident in question (though I do remember one with Dunn after MacKanin took over this year--though Dunn handled his objections very appropriately), but Griffey was fairly bitter about the position change all season.  He usually says something like “I’m just player #3.” It’s clear he doesn’t agree with the idea, and that’s among the reasons I’ve long suspected that he’d accept a trade, especially to a contender that would let him play in the outfield (I severely doubt he’d go to DH willingly).  Unfortunately, the Reds won’t deal him because he’s supposedly too much of a draw, especially as he inches toward 600 hr’s. 

Winning’s a pretty good draw too…
-j

Hmm… If the Twins tried to get him to DH, Griffey would be thinking “they just lost Torii Hunter, they need me in center...”

Cordero: 4/46, paying for 3.0 WAR.

A 3.0 WAR implies a relief pitcher (72 IP) with a win% of .700.  This is calculated as:
(.570 - .470) * 1 = .100
+(.708 - .570) * 2 = .260

where the “* 2” is the LI.  The total wins is +.376 per 9 IP, or a total of 3.0 WAR.

In post 60, I had Mo as a .700 pitcher (3.0 WAR).  I can’t have Cordero that high.  I would say .625 or .650 pitcher, which implies a WAR of close to 2.0, or a salary of 3/23. 

A great reliever, obviously.  Since 2002, his WPA has been +9.3 wins, with an LI of 1.9, on 407 IP, which means his unleveraged WPA is .610.

Relievers are incredibly overvalued, and this is yet another case.  I’d like to know what was his second-best offer on the table.

I’m also not too concerned about his age.  If he’s throwing fastballs with movement at 96, he’s got a “young arm”.  It’s like Mo.  I would guess that all old starting pitchers who used to have plus fastballs should be turned into relievers.

#173, FWIW, the Brewers supposedly offered him 4/42 according to Rosenthal.

http://msn.foxsports.com/mlb/story/7480854

Their ace relief model is broken.  They are giving the entire leverage to the reliever, when the Guy model works out better.

***

Santana:

Since 2004, he’s had am unweighted WPA/LI of almost +17 wins in 912 IP, or a .667 pitcher, about as good as it gets.  Last year, he gave up lots of HR (his FB rate was constant, but his HR/FB shot up).  I’m presuming this is not due to a change in his approach or mechanics, but this is the kind of thing you’d like to go to the pitch-by-pitch to see if there’s a change in true talent, or purely a sampling issue.

I think it’s fair to call him a 6 or 7 WAR pitcher.  He’s also at or near his peak, so don’t expect a dropoff.  You might expect even improvement.  A 6 WAR FREE AGENT pitcher is worth 27MM, and a 7 WAR pitcher is worth 31 MM.  Santana makes 13.25 according to Cot’s:
http://mlbcontracts.blogspot.com/2005/01/minnesota-twins_17.html

So, the free agent value of Santana is about 14MM to 18MM.  That’s how much discount you get from him, compared to a free agent of equal talent.

His 13.25 however is pretty much exactly what he is worth in terms of generating revenue.  If all players were free agents, then all players would earn 2.3MM per win, or in the case of Santana, have a free-est market value of 14 to 16MM.  But, that’s not the baseball market.

Since he is saving you 14 to 18MM, any trade in return must give up property at those levels of savings.

See, for the Twins, they are properly paying for him talent for 2008.  But, for the other 29 teams, that compete for free agent pitchers, they see someone very undervalued, and are willing to pay for it with players.  It’s a very fascinating situation for the Twins.

http://www.insidethebook.com/ee/index.php/site/comments/do_pitchers_today_have_it_better_because_of_medicine_and_technology/

This link is fairly important, as Santana and Buehrle are the same age, so we can apply what we learned from that thread, notably The Rule of 10.

We expect Santana’s IP to drop 10% every year.  So, this is our IP for the next 4 years 207, 188, 171, 156.

If he’s a .670 pitcher today, then he gets these for the next 4 years: .660, .650, .640, .630.

This gives us these WAR: 6.4, 5.6, 4.9, 4.3.

We use these marginal $/win: 4.4, 4.84, 5.32, 5.86.

(Note: gotta love The Rule of 10!)

Over the next 4 years, that gives us a total of 4/108.

If we keep going, that’s 5/133, 6/157.

These salaries are consistent with someone being evaluated as a 6 WAR pitcher, aging/declining/injuring at 0.5.

Basically look for a 26MM per year offer, with number of years up for grabs.

The Twins also have another strong potential trading chip I don’t see mentioned much: Joe Nathan, signed for one more year at around $6-$7 million. As its been noted, relievers of Nathan’s calibur (and he might be the best in the game right now) are highly coveted. If teams considering Santana get their collective sphincters too tight that the Twins don’t feel it’s worth it at this point to trade him, they could still look to deal Nathan and see if they can supplement their sorry lineup with a player or two. And though the Twins are by no means sabermetrically sound, one thing they’ve been consistently excellent at over the past decade is building high-quality bullpens out of relative no-names. With their collection of young arms, I think they’d be fine.

Of course, if they do trade Santana, then there’d be presumably little point in NOT dealing Nathan, as well. The Twins could come out of this offseason an entirely different entity.

Regards to #110, I understand what you’re trying to accomplish by presenting WPA/LI, and it does make sense and would be nice to look at with MGL’s lwts, but couldn’t you only use WPA/LI for the previous year(and not regress past years like MGL does with lwts)?  Or am I wrong?  In other words, is WPA/LI like regular WPA in that it measures value retrospectively, and therefore does not have the predictive value that MGL’s lwts has?

WPA/LI is very much like MGL’s offensive lwts (hitting, basestealing, excluding baserunning, not park-, nor league- adjusted).

If you go to Fangraphs and sort by WPA/LI, you get this list:
+6.4 ARod
+6.3 Ortiz
+5.6 Holliday
+5.2 Ordonez
+5.2 Fielder
+4.9 Utley
+4.9 Wright
+4.7 Chipper
+4.6 Carlos Pena
+4.4 Pujols
Next highest is +3.8

I would bet that MGL’s list would be extremely close to this.

The only difference in approach is that MGL gives no weight whatsoever to how a player performs with men on base or in a close situation, while WPA/LI reweights every event to that situation.  It does NOT give higher weight to every event in high LI situations.  It reweights all the events.

So, if you can imagine that a walk gets a weight of 0.7 and a HR gets a weight of 2.0 in a typical situation (which is what wOBA does), then in a particular situation, the walk could come in at a value of 0.5 and a HR at a value of 2.5 in some other situation, or 0.9 and 1.2 in some other situation.

It recalibrates the weights, while maintaining an overall average of .340 (which is your league OBP basically) for each and every game state.

It treats it as a fact that players will perform differently with man on 3B and less than 2 outs, something which we know to be true.  The question is whether the sample data we collect in these cases is enough to offset the small sample size, that we wouldn’t be better off considering that all players perform the same in all game states.

Clearly, this is not true at a certain level (say 10,000 PA).  Clearly, mgl’s approach is better at a certain level (say 30 PA).  What’s unknown is at what point they are break-even (and you need to consider 50% of each).

Tango, I don’t think you answered at least part of Mike’s question. He asked whether you can use ONE YEAR of WPA/LI to make projections AND whether you can NOT do a regression (regress those stats toward some mean).  The answer, Mike, is that you can do anything you want.

Seriously, the answer and issue are complicated.  You can use one year’s worth (or any time period) of any stat to use as a basis for a projection, but that is generally not going to be nearly as accurate as using multi-year stats.  You want to use as many stats as you possibly can to estimate true talent (which is essentially the same as a projection), in order to reduce sample error.  As you know, if a player has one PA, regardless of how he does, that result is not a good predictor (in fact, almost completely worthless, right?) of future performance.  So too with 10 PA, right.

At what point (how many PA, if we are using PA) DOES past performance become a good indicator?  There is no magic number.  The more the better, because eventually there will be NO sample error and a player’s past performance (at an infinite number of PA) will be EXACTLY equal to his average true talent over that infinite period of time.

However, in baseball, we have a problem that true talent changes over time and it is more likely that recent past performance reflects current or future true talent than not-so-recent performance.  So we have to establish a balance between those 2 things - using a large sample to reduce sample error (random fluctuation or so-called short-term “luck") and using more recent performance, since it more reflects current and future true talent levels.  The way we do that is to weight performance by recency.  The more recent the performance, the more weight it gets, per PA.

Regression toward the mean is sort of a separate issue.  We use regression toward the mean in these kinds of things to “make up for” sample error, using information we know about a player other than his actual performance.  Again, if a player has 1 PA, that does not help us much in predicting his future performance, but believe it or not, if we knew nothing about baseball and baseball performance, our best guess for that player’s future performance WOULD BE his BA (or whatever we are measuring) in thst one PA.  If he makes an out, we call him a .000 hitter and that is his projection.  If he gets a hit, we call him a 1.000 hitter.

But, we DO know something about baseball players and baseball performance to make those projections unlikely.  We know that most players hit around .200 to .350 in true talent, with an average (mean) of around .260 (or whatever the numbers are).  So, armed with that information, regardless of what a player does in that one PA, we call him a .260 hitter.  THAT is regression toward the mean.  For one PA, we regress the actual .000 or 1.000 BA almost 100% toward the mean.  If he had 500 PA, we might regress 50% (or whatever it is) toward the mean.

How much we regress a stat towards the mean depends on two things:  One, how many trials (like PA) it is based on, and two, how much spread of true talent there is in the population of players that our one player belongs to.  The greater the spread, the less we regress.  Why is that?  Simple.  If our player hits .400 in 500 PA, and almost all players are actually true .250 to .270 hitters (which would mean a small spread in talent right, right?), then it is extrememly likely that our .400 hitter just got lucky in those 500 PA. After all, almost no one in our population is a high BA player.

On the other hand, if all hitters were between true .100 to .400 hitters (which they aren’t of course), which is a much larger spread in true talent, then there is a decent chance that our .400 hitter in 500 PA is really a high BA hitter, so we don’t want to regress that .400 toward the population mean of .260 very much when doing a projection or estimating our player’s “true” BA.  Basically, “regression toward the mean” in these kinds of calculations is short-hand for, “What are the chances that he is a true .200 hitter who got lucky in 500 PA, and what are the chances that he is a true .210 hitter who got lucky..., etc.” This is called Bayesian probability.  Bayesian probablity questions usually involve two sets of probability distributions.  In this case, they are the binomial one for 500 PA (where p=.260) and the distribution of hitters in MLB (e.g., 30% are true
.260, 10% are true .220 or .290, etc.).  Bayesian problems also usually answer the general question, “What is the answer to this probability question, GIVEN a set of separate probabilities that we know?” In this case, it is, “What is the best estimate of the true BA for a player who hits .400 (or whatever) in 500 PA (or whatever PA), given that x% of MLB hitters (who get 500 PA in the same time as our player) are true .210 hitters, y% are true .200 hitters, z% are true .300 hitters, etc.

Finally, whether WPA/LI is a better predictor of future performance (namely WPA/LI, which is indeed what we care about in terms of future value - actually we care about WPA, which includes clutch/leverage hitting, but we are pretty sure that pure WPA by itself is NOT very predictable from past WPA by itself) than a context neutral stat like lwts, is debatable.  We KNOW that players perform differently in different bases/outs state (and WPA/LI captures that), but we DO NOT know whether those differences are appreciably different among players (we can assume that there is SOME talent/difference among players) to use past WPA/LI to predict future WPA/LI.  I doubt that it is.  But I have not seen the work done to answer that question one way or another.  And that work can be easily done (by someone who knows how to do it of course).

Now, as with pure clutch hitting (WPA-lwts, essentally, which icludes clutch with respect to bases/outs as well as leverage), we KNOW that the correct thing is to take context neutral stats like lwts and then ADD in a regressed form of WPA/LI minus lwts.  But how much to regress the WPA/LI over and above the lwts, we don’t yet know.  My guess is that it is fairly small, but certainly not as small as pure clutch.  And a good question is that given the regression amount for WPA/LI minus lwts, is it better to use WPA/LI to predict future WPA/LI or is it better to use lwts (a context-neutral stat) to predict WPA/LI, if those were your only choices?  My guess is that it is the latter, but it could be the other way around.

Mike, if you don’t understand the last paragraph, don’t worry about it.

Basically, in your post, you were convoluting several related issues.  When you say, “In other words,” it should not have been “in other words,” as they were related, but not the same, issues. 

Hope that helps.

Papelbon:

(cross-posted to http://sonsofsamhorn.net/index.php?showtopic=25754 )

Papelbon has 2yr, 64 days service, through 2007.

There are 137 players with at least 2 years, but less than 3.  The 22 (I think it’s the top one-sixth) are classified as super twos (Roger Clemens rule).  That puts the line at 135 days.  However, I see that the guy below that line is Weeks, and he’s part of Cot’s list.  The guy below Weeks (Kelly Johnson) is not on Cot’s list.  So, I guess the line is 131 days.
http://mlbcontracts.blogspot.com/2000/05/2008-arbitration-eligibles.html

In any case, Papelbon is way below that line.  His first arb year is next year (he’ll be 3+ years), and those guys earn 40% of their free agent value.  His other two arb years (4+ and 5+ years of service) are worth 60% and 80% of free agent value.

Presuming that his free agent dollar value this year is 12MM, and salaries increase at 10%, and he maintains his performance (no injuries), then the Redsox should sign him for a 4/27 deal.  And I think that’s being generous.  Pitchers do get hurt.  If we put in some potential for that, then I estimate a 4/20 deal to be fair.

Mets and WAS had a fairly big trade of course.  Milledge for Schneider and Church.

I think that most Mets’ fans (and other people) have lambasted Minaya for this trade.

The “analyses” I have seen (not many) so far, as usual, completely omit each player’s respective salaries and situation (MLB service time, which impacts future salary of course).  It is amazing how nearly everyone, including the media, analyzes trades without considering contracts.

Obviously, as most people on THIS board know, there is NO way to analyze a trade without looking at each player’s contract.  I don’t really need to explain why here (but I will).  In baseball, you don’t trade players, like, “I’ll give you my bike for your Playstation.” You trade players and their contracts.  The value of each player in the trade is NOT the win value of the player, but the win value of the player MINUS their salary, or the “net equity” of the player.

As Tango likes to say, it is like trading houses with mortgages.  Everyone knows that you can’t evaluate a “house trade” without looking at the mortgages.  Somehow people just don’t seem to realize that baseball trades entail EXACTLY the same thing.

Just like if I give you my 2 mil house (fair maket value) with a 2 mil mortgage (with a prevailing interest rate) for your free and clear junky old $200,000 house, I just got a steal (I earned $200,000 in equity), if you traded A-Rod for Milledge, you would get a steal, and if you traded Milledge for Jeter, you would have gotten ripped off to no end, based on this model.

Anyway…

Schneider, according to my projections, is around 1.5 WAR.  He can’t hit for beans (which I always think is amazing for a switch hitter, let alone a physically large one), but is still a great defensive catcher.  I give him around .5 wins in defense and 1 win above replacement in hitting (for a catcher of course).  He is getting 4.5 m per, I think, which is probably a little low (as a FA).  I think he has a couple mil per year in equity.  So simply acquiring him at all is a pretty good deal.  Of course the Mets now have 3 catchers, and I don’t know what they are going to do with them.  Plus, for some reason, they have never liked Castro much, who is an excellent hitter for a catcher.  According to his numbers, he whould be a starter on many teams and in the league in general.  Maybe he can’t play that many games for some reason, IDK.

Anyway, Church and Milledge are very similar (now), other than age, according to my projections.  I have both of them projected at around 2.5 wins above replacement on the corners.  Church has around 1 year of service and Milledge has almost none and they both make around the minimum.  IOW, they both have around 12 mil in equity per year in them.  They are both extremely valuable players!

Now, the difference is that Church is 30 and Milledge is only 22.  Each year, Church is going to get a little to a lot worse and Milledge is going to get A LOT better (presumably of course).  So, over the next 5 to 6 years, the time periods in which the respective teams will control both players, Milledge is A LOT more valuable.  In fact, given Milledge’s age, he really is projected to be a star or superstar player.

In addition, Milledge is and was considered a star prospect, while for some reason Church is still a AAAA player at the age of 29.  Given my very good projection for Church, there MUST be something about him that the teams/scouts do not like.  So even though they both project about the same next year, I think that you have to take the Church projection with a little bit of a grain of salt (although I definitely think he is a way underrated player for some reason), and give the edge to Milledge.

So, if you were to trade Church for Milledge straight up, it would be a steal for the team that gets Milledge.  Heck, in 4 years, Church is going to be near replacement and Milledge will be a near superstar (again, presumably).

Given that the Mets are getting a little bit of equity in Schneider, that mitigates the deal for the Mets a little.

So, all in all, I think it is a pretty bad deal for the Mets, but not horrible, if only because of Schneider and because most people probably underrate Church by quite a bit (how many 2.5 WAR players are there that most people have never heard of and hardly play?).

I welcome other analyses.

Church is a “super 2”, meaning he has 2+ years of service AND arb rights right now.  He will get 20% of his free-agent value this year, 40% next year, 60% the year after that, and finally 80% in his last arb year.

Milledge is 1.004, meaning he will only have arb rights in TWO years (i.e, he plays for near minimum in 2008 and 2009).  Lastings will get 40% of his free agent value in 2010, and 60% in 2011.  A fantastic opportunity to “try before you buy” for two years, like leasing a car for 9$ a month.

Fans think of both fielders as even fielding-wise.

Schneider is, from the Mets perspective, the “equalizer”.  Hard to believe that there wasn’t 1 out of 28 teams willing to offer more.

With Church, I would susepct that his FA value, according to teams/arbitrators will be less than his FA value according to my projections.  (BTW, I am using 5 mil per win.)

“More,” to or for whom?  To the Nats for Schneider?  It looks like that they got a great deal for Schneider and Church.  Or do you mean more for Milledge?

Right, more for Milledge.  Dude has almost no service time, and yet can be a regular OF.  That’s a great “try before you buy” player.  For two years, you pay him almost nothing.

And I think that’s what has caused most of the consternation, trying to fathom that this was really the best deal.

I mean, this is a guy who was brought up consistently last offseason in the “Maybe the Marlins should cut bait on Dontrelle before it’s too late” rumors.  Someone who was thought to be one of the apples of Billy Beane’s eye in previous years. 

So for this to be the ultimate outcome, after such a short showing with the Mets, I think a lot of people are trying to decide whether to believe that this is really where his value is seen, or whether Minaya should have gotten more.

By the way, Schneider is a LHB not a switch hitter, not that it changes the analysis.
http://sports.yahoo.com/mlb/players/6480

I agree with MGL in that people completely overlook the role of contracts in trades. Every trade I have seen proposed for Johan Santana would be a good one for the Twins in that they will receive more (in some cases, way more) in surplus value than they would be giving out. Tango calculated Santana as a $28M player for 2008 (the only year he has left under contract) and he makes $13M. So, he has $15M of surplus value which I reckon is less than someone like Clay Buchholz or Phil Hughes would have by themselves.

If Hughes or Chamberlain or (Tulo, Braun, Pedroia) other top flight rookies/sophs are even just average overall players, their free agent value is already at 9MM, even though they cost like 500K.  So, after their first two years alone, you are getting some 17MM in surplus free agent value.

Then for their 3 or 4 arb years?  That’s tons of surplus value

Giving up Santana, one on one for those guys is already a good deal for the Twins.

This “surplus” value can be found in hockey, right after the lockout.

They decided after the lockout to hold a lottery for the first over draft pick.  That pick, Sidney Crosby, was highly anticipated (at the Lindros, Lemieux, Gretzky level).  The agreement also had a hard cap on rookie salaries.  In effect, any team drafting him would be saving themselves 10MM every year.  That is, he’d produce 11MM of value for 1MM for the first 3 years.

At the same time, the Penguins were put on the block, and ready to be sold.  A price was already agreed upon.  But, the Penguins won the lottery, and the rights to draft Crosby (this is bigger than drafting Junior Griffey).  So, you’ve now got 30MM of surplus value here.  Since the player is not getting it, guess who?  That’s right, the franchise value.  The Penguins sold a minority stake at a higher franchise value, strictly based on the extra value that Crosby generates that’s not paid for.

The MLB system is FANTASTIC for the A’s, Twins, and Marlins and other small market teams, if they can game the system.  Their 4th and 5th year players are sought after like crazy, and the other teams’ prospects are undervalued (my guess is that those teams think that since they’re not paying much for them, they’re not as valuable a loss).  The small-market teams can continue to pay pennies and quarters on the dollar that rich teams pay dollar-for-dollar.

I think whatever team signs Santana has a pretty good chance at getting him under the $28 million he’s supposed to be worth.  No pitcher has ever broken the 20 million dollar barrier.  The rumors of his asking price make me think a 6 year deal a bit over 20 million would get him signed.  Whoever trades for him could be getting a lot of surplus value in an extension.

So I can see why a team would be willing to part with more than 15 million worth of young talent to trade for him.

Yeah, but if he is worth 28 mil next year, what is he worth for a 6 year deal given what we know about how pitchers’ average IP decline each year and how their rate performance declines each year.  Tango’s chart may apply to position players.  I don’t think it applies to pitchers.

I have Santana as exactly 6 WAR in 210 IP.  That is close to 30mm in FA value for 2008.  I am going to say that his expected IP goes down 15% per year and his RA goes down .1 per year.  That is a loss of .87 wins in year 2!  His loss in wins per year will actually decrease as his IP decrease at a 15% rate, so we’ll call it .75 wins per year loss.

year 1: 6.0
year 2: 5.25
year 3: 4.5
year 4: 3.75
year 5: 3.00
year 6: 2.25

At a 10% inflation rate and 5 mil per win, that is:

year 1: 30mm
year 2: 28.9
year 3: 27.2
year 4: 25.0
year 5: 21.9
year 6: 18.1

That is a 6/151.  If he signs for 6 years, I think he’ll sign for something like that.  I really do.  And using 5mm per win is the high end I think.  I am using 1.2 runs per 9 worse than the league average RA for a replacement pitcher and I am using 1.2 runs better than league average RA for Santana in year 1, and I am assuming 7 IP per game for him and for the replacement pitcher (which is where you run into problems figuring Santana’s WAR - a replacement pitcher is only going to go 5 IP per game and then get replaced by replacement relievers who are only .2 or .3 runs worse than league average rather than 1.2), and 210 IP for Santana in year 1 (with a 15% decline in IP each year and a .1 decline per year in RA).

Peavy:

He will have reached his 6 years of service in the middle of 2008, meaning he’d otherwise be a free agent heading into 2009.  He’s signed for 11MM, which is a fantastic deal for the Padres of course.

They are talking about a 3-yr extension of around 3/52 for 2010-212.

Paying 3/52, in the future, for 3.4 WAR starting in 2010.  This would be the same as saying he’s 4.4 WAR today, and dropping him by 0.5 wins every year.

4.4 WAR today for say 207 IP means that he’s a .570 pitcher.

I can’t believe Peavy would sign such a deal.

***

By the way, presuming he’s a 4.4 WAR pitcher (which is too low), means that he’s worth 39MM for 2008-2009.  He’s being paid 17.5MM.  He has a free agent surplus value of 21.5MM, which would make him one of the most attractive players on the market, even if you think he’s only a .570 pitcher.

Padres are going to have fantastically underpaid for Peavy for his career.

The Santana/replacement discussion has been moved to its own thread:
http://www.insidethebook.com/ee/index.php/site/comments/starters_going_deep/

***

Re: 190

As for the notion about breaking the 20MM barrier with a 25MM deal, isn’t that what ARod did 7 years ago?  Santana is to the rest of the league what ARod was to the rest of the league.

I think I called for 6/150 deal as well.

It’s just too bad he didn’t have his typical great year last year, what with all those HR. 

Jose Guillen: 3/36, paying for 3.0 WAR.

You’re kidding?  Mariners rejected an option for 1 yr at 8 or 9 MM (implying they didn’t think he was a 2 WAR player).  Couldn’t the Royals have asked the Mariners to sign him, and give him some low-level prospect in a sign-and-trade?

Guillen is a bit above average hitter for RF, and a bit below average fielder for RF.  He is, basically, average.

(Fielding-wise, Fans and Dewan see him as average to a shade above.  UZR sees him as way below.)

He should have been signed to a 8 or 9 MM per, at 3 years or less.

This is either a big black strike against my salary chart, or the Royals.

I have Guillen (32 years old in 08!) as barely 1.5 WAR.  Even using my 5mm per win, that is 8 mil as Tango says.  Bad signing!  But you expect bad signings from bad teams of corner OF’s and 1st basemen, as they fail to understand positional value.  These players, even as average hitters for their position, look like star hitters, especially if they happen to have good garbage stats and/or have lots of HR.

My bad rating for Guillen is predicated on almost -1 win in defense.  However, even if he is average defensively (which for a 32 year old would have required him to have been very good at one time), he is only a 2 WAR or 3/30 at best.

He was!… according to you (at one time):

http://www.tangotiger.net/UZR0003.html

Top RF in MLB, +22 per 162, on 187 games.

Regressed, he was +11 runs in RF:
http://www.tangotiger.net/UZR9903TT.html

I have revised my UZR several times over the last few years, but here is is UZR last 4 years:

07 -21
06 -3
05 +11
04 -19

03 +2
02 -10 (50 games)
01 +18 (only 38 games)
00 +15 (82 games)

An oft-injured 32 yo OF’er is not likely to have much of a defensive rating, unless he was a star defensive player at one time, or aged extremely well, but you never know.

Maybe there is something about RF in SEA that depresses UZR, I don’t know.  I never liked Ichiro there.  Then again, I don’t like him in CF in SEA either.  I’ll have to look at my road values for the SEA OFers.

"Couldn’t the Royals have asked the Mariners to sign him, and give him some low-level prospect in a sign-and-trade?”

If the M’s picked up his option, Guillen still had the right to decline it and become a free agent which he certainly would have done.

David Riske: 3/13 paying for a little < 1.5 WAR;
incentives could bring the total to $20MM (about 2.0 WAR) and there’s a 2011 option at $4.75MM.

I’ve gotten lost with all the discussions about replacement-level pitching, so I won’t even bother trying.

For non-closers, the aging aspect is a bit confusing (it’s closer to a .20 wins drop per year).

In the above case (3/13), they are paying for 2.50 WAR (total) over the next 3 years.

That means they are paying him to be a .590 pitcher in 2008, with 72 innings, and then dropping by .010 wins each year and 10% innings each year. 

If you make him a .550 pitcher with 72 innings, he’s a 3/8 pitcher. 

He’s an interesting case.  There are 55 active relievers since 2003 with at least 250 IP, but less than 50 saves.  He’s 13th in OPS+, along with Kyle Farnsworth, Latroy Hawkins, Timlin, Mota.  He’s 6th in ERA+.

It’s a tough call deciding if he’s closer to be a .600 reliever or .550.  Basically, each 10 win percentage points costs you a million$ in evaluation.

A study of salaries for non-closing relievers would be fascinating.

I have Riske as .5 runs per 9 better than a repl. reliever in 08.  Where repl. rel = .3 runs worse than league average rpg.

I guess a better way to say that is Riske is .2 runs better than league average rpg.

Giving him 8 “complete games” and that puts him at +4 runs above replacement, which is 0.4 WAR.  At that level, he is fantastically overpaid.  Over the next 3 years, he’d be like a 0.7 WAR pitcher total, meaning a 3yr/5MM pitcher.

Andruw: 2/36, paying for 4.0 WAR.

His offense would be +1.5 wins, his fielding +1.0, position is +0.5, and aging is -0.5, for a total of +2.5 WAA, or +4.5 WAR.

Fielding: Fans have him at +1 win, Dewan as the best CF in baseball (+1.5), and MGL’s 03-mid07 as league-average (0).

Dodgers likely got a good deal.  I think it’s likely that teams and players don’t follow a Marcel forecasting model (5/4/3), but more like a 4/2/1 model.  A 4/2/1 would give Andruw +1 win as a hitter.

Andruw is not old enough (dude will be 31 next year) for me to believe that a 4/2/1 model would be appropriate.  For an older player, like Frank Thomas or Mike Piazza, it might make more sense.  The reason we weight the more recent season more is because that’s what his body and skillset is closest to.  The older you are, the more you care about recent performance (and the younger you are too).  If you are in the 24-34 age group, 5/4/3 is likely very appropriate.  If you are 18, you much more care about your age 17 performance that your age 15 performance.

Now, for the 4/2/1 to apply to Andruw, as opposed to 5/4/3, you have to believe that he underwent something fundamentally different, like he aged really badly, he lost two steps, or his mind has turned to mush.  Something drastic, almost like he’s really 38 years old.

But, you MUST make that decision INDEPENDENT of his actual performance.  You can’t try to explain his performance results (output) with a possible input change.

Applying a 4/2/1 to Andruw based on B-ref’s batting wins, I get 0.4 wins over average (including regression won’t change it much).

A 5/4/3 gives him 1.0, with regression its 0.9.

And that’s in the weaker league.  An AL team would probably see him as an average hitter.

I have Andruw projected at a WAA in offense and zero at defense, as Tango says I would.  I’ll grant him .5 in de.  That puts him at 1.5 plus position or 2 WAA or 3.5 to 4 AR.

So far, we have not been adjusting for league, but Rally is right.  We probably should be giving hitters and pitchers -.2 win for NL and +.2 for AL or something like that (is that too much?).

So we’ll make AJ 3.5 WAR.  2/36 is not horrible, but not great either considering that he had such a poor 07 (-11!) and teams usually discount a player because of that.

I am not a bog fan of AJ.  Tango (or someone), what is his WPA/LI or WPA as compared to what it should be?  My guess is that it is less.  He seems to me to be the type of player who does not adjust well for the situation. He just seems to always swing as hard as he can.  Doesn’t run out ground balls either.  I expect that from a DH type, but not from a supposedly fast CFer.  Interestingly, he has a -1 projected lwts in “GDP tendency” which suggests that a player is either not fast or doesn’t run out ground balls, even in DP situations.

They discounted him on the years instead of on the dollars.  I’ve got Hunter as a slightly better player, but from a team’s perspective I’d much rather have the Andruw deal.  Odds are against Torii being worth 18 million when he’s 35-37.

MGL gave him a +2.0 WAA, compared to my +2.5 WAA.  The difference being in his fielding (mgl no like).

In terms of repl, you add +2.25 for me, which puts him at +4.75 per 162 G.  If we give him 150 G, that’s 4.4 WAR for me.

For MGL, his repl level is a bit higher, whereby you add +1.944, or he’s at +3.944 WAR per 162.  At 150, that’s 3.65 WAR.

My WAR$ cost 4.4 each, while they cost MGL 5.0 each.  It’s 18MM for MGL and 19MM for me.  Basically, we are in agreement as to his cost.

***

Yes, I think I said we should do +0.2 and -0.2 for the league adjustments.  I think I said we should treat the avg AL team as being .525 and the average NL team as being .475.  That means each AL team needs an extra 4 wins allocated to its players (2.3 to nonpitchers and 1.7 to pitchers).  So, 2.3 wins per 9 fulltime players means 0.26 wins per 162 G for nonpitchers and .010 wins per 9 IP for pitchers.  And of course, the opposite for NL.  (Presumes each league equally good/bad in their hitting/pitching.)

I’ve definitely been lazy in not applying these adjustments.  Andruw’s 4.4 WAR would come down to 4.2 WAR, which is basically what he signed for.  Good job Rally.

***

His WPA/LI has been great.  If we look at his Palmer Linear Weights (under BtnWins at b-r.com), since 2002, he’s been: +8.3 wins
And at Fangraphs, he’s been: +11.5 wins

http://www.baseball-reference.com/j/jonesan01.shtml

http://www.fangraphs.com/statss.aspx?playerid=96&position=OF

So, that’s a 3.2 win difference over 6 years, which is substantial.  That is, he performs differently enough in game situations (i.e., doesn’t strike out when he shouldn’t, gets a walk when it’s more valuable, etc), that he has an enormous difference.

Remember, WPA/LI, as opposed to WPA, weights each and every PA equally.  You do not get an extra benefit in performing well when the game is on the line.

I had Hunter at 4 WAR and Andruw at 4.5 WAR.  Once you apply the league-adjustment to both, you end up with the same. 

Going year-to-year on a free agent contract is always preferred.

Hunter’s 5/90 is really being paid like this, in those year’s dollars:
Years Salary
2008 $19.63
2009 $19.13
2010 $18.35
2011 $17.21
2012 $15.68

I know he’s getting a straight 18MM per, but this is basically how much he’s earning.

And, if we put this in 2008 baseball dollars, he’s getting this much:

Years Discounted
2008 $19.63
2009 $17.43
2010 $15.23
2011 $13.03
2012 $10.83

That is, you are paying Torii 76MM in 2008, and zero in 2009 through 2012.  Or 20,19,18,17,16.  It’s the same thing.

I don’t like the “he won’t be worth 18MM when he’s 37 talk”.  He’s basically deferring part of his 2008 salary into 2012.

Remember, in 2012, 18MM then will be 12MM today.

Yes, I agree that the talk about “not being worth X when he is Y years old” is silly.  Of course A-Rod is not going to be worth 25 mil when he is 42, even in current dollars (maybe 10 mil?).  I assume there is like an 80% chance he is out of baseball by then.

It is irrelevant what you pay a player in the future. Think about all salary packages like an annuity - like he just won a “million dollar” lottery that pays out 50,000 a year for 20 years.  Hunter’s salary is simply 90 mil (or 76 mil in current dollars) and he plays for “up to” 5 years, or, alternately, 5 years, or when he gets terminally injured, retires, or gets released, whichever comes first.  Same for Andruw and everyone else who signs a mulit-year deal.

In fact, maybe we should just start listing player’s salaries in 2008 dollars and that is it.  And as an aside, we can list “for up to X years.” Hunter just signed for 76 mil.  Period.  Oh, and he has to play for up to 5 years with the Angels.  If he quits, of course he has to forfeit part of that 76 mil.

I like it!

I’m interested in going back to this:

If Hughes or Chamberlain or (Tulo, Braun, Pedroia) other top flight rookies/sophs are even just average overall players, their free agent value is already at 9MM, even though they cost like 500K.  So, after their first two years alone, you are getting some 17MM in surplus free agent value.

Giving up Santana, one on one for those guys is already a good deal for the Twins.

Is there a GM in baseball that would trade Santana straight-up for an average rookie, outside of a salary dump (in which case the $17M likely wouldn’t be going toward improving the team, but improving the owner’s retirement fund)?  I simply cannot imagine it happening.

Is every GM wrong, or is the model being used here not reflective of reality?

Two centerfielders worth about 4 wins each.  Both over 30, presunably in decline.  Both make 18 million per year.  Excuse me if I wish my team signed the one for two years.  Using $4.4 and 10% each year, declining by 0.5 wins per year, Hunter is worth:

2008 17.6
2009 16.9
2010 16.0
2011 14.6
2012 12.9

Under these assumptions inflation is not keeping up with his decline.  Angels overpay by 12 million, Dodgers only by 1.5.

As I said:

“Going year-to-year on a free agent contract is always preferred. “

And I think I called Hunter a 5/80 player, not the 5/90 he signed for.  So, no disagreement here.

***

Yes, Santana should be traded straight up for an average rookie, if and only if, you are giving up 1 year of Santana, and you indeed do have an average rookie.  What would you rather have for 5 years:
Santana, 18-9
replacement, 11-16
replacement, 11-16
replacement, 11-16
replacement, 11-16
Cost in today’s dollars: 16MM for 62 wins - 73 losses

Rookie, 13-14
Rookie, 14-13
Rookie, 14-13
Rookie, 15-12
Rookie, 15-12
Cost in today’s dollars: 16MM for 71 wins - 64 losses

You get 9 wins spread out over 5 years.  If you are not competing this year, better to give up on Santana now.

Let us not forget that while players expect to give a discout for the security of a longer-term contract, and rightfully so, so do teams expect a discount for a longer-term contract.  I don’t think Tango’s salary model takes that into consideration.  It should, although you would have to either guess at what that should be or try and figure it out from the contracts already given over many years.

Not to mention the fact that a shorter term contracts is more optimal in terms of giving you flexibility later on in case your needs change.  That should reduce the value (for a team) of a long-term contract as well. I would not be surprised if we could say that a team should expect to pay 5% per year less on a contract for a position player and 10% for a pitcher.  If you did it that way, you would have to increase the $ per win for year 1 since the dollar values are based on all contracts, including long-term ones.

I’m not sure there is much discount on free agent contracts.  However, I’m quite certain there are non-free agents.  Chase Utley, Joe Mauer and the other signings in 2007, which my model missed bigtime on, are some evidence to that.

Basically, if you have not earned alot of money in MLB, you want to cash in your chips, and are willing to sacrifice alot of your future earnings.  If you have already earned alot of money, then, you are not going to give anyone a discount, since you have 30 suitors.  One of them will see things your way.

Well, I think it is that we just don’t SEE the discount because every good FA gets multi-year contracts generally.  IF they were to entertain a one-year contract, say, it would have to be a lot higher.  It has to be that way.  Let’s say Hunter gets offered 5/90, which he did.  How much do you think it would take for him to accept a one-year offer?  Certainly nothing less than 25.  It just doesn’t happen that way that often or we don’t hear about it.  A. Jones may be a good example, since his 2 year contract is quite unusual for a very good established player at 31 years old.  He probably could have gotten a 4/50 or something like that if wanted, no?  He may have even been offered that by someone, although a team would probably know that he would have turned it down.  But surely he would have had to consider something like a 4/60.

You are right that we don’t “see” it.

A guy who signs 2/36 (implying a WAR of 4.08 in 2008, 3.58 in 2009, with a 4.4MM per win in 2008 and 4.84MM per win in 2009), has these equivalencies if his performance and costs per win continue on their path:
1/18, 2/36, 3/53, 4/68, 5/82, 6/94

Hunter’s 5/90 implies the following:
1/20, 2/39, 3/57, 4/74, 5/90, 6/104

What you are suggesting is that Andruw got a premium and/or Torii got a discount.  How much of a discount could Torii have given?  Let’s say that we go crazy, and say that Torii gave a 3% discount for 2009, 6% for 2010, 9% for 2011, and so on. 

Under that scenario, Torii’s 2008 expected WAR goes from 4.37 to 4.57.  Here’s what his salary should be, if discounting of 3% applies:
1/20, 2/40, 3/58, 4/75, 5/90, 6/103

As you can see, not much really changes.  If we presume he got 5/90 with no discounting, then he would have signed a 3/57 deal.  If he got 5/90 with a discount, he would have signed a 3/58 deal.

What if we make the discount at 5% in 09, 10% in 2010, etc?  Under that scenario, he would have had an implied WAR of 4.7 for 2008, which is starting to get too high.  Anyway, this would have meant a 1/21, 2/41, 3/59 deal.  Again, not much difference really.

I really don’t think that teams/players think of discounting on long-term deals, for free agents.  They almost definitely do with pre-FA.

Sorry if this is a double post, but I think teams and free agent players definitely give up salary for years.  Just wrote an article at THT about it.  They may not “think” about it, but it’s sewn into the fabric of the market at this point.

I agree that we don’t see it with your model, but I kind of doubt your model is capturing everything that happens in the market.

I don’t think Torii Hunter would view 1/20 as equal to what he got from the Angels.  If another team had offered 1/21 I don’t think he would have seen it as a better offer, or even considered it for a second.

But the Angels might view them that way.

Not sure if this makes any sense, but use a slight discount from the teams’ perspective, as in post #217, and from the player’s perspective a steep discount, for Torii start at 1/27 ... then whatever it takes to get to 5/90. 

Where the two graphs meet you have the deal in dollars and years.  Again, just a brainstorm, not sure if it makes sense.

If Torii’s 5/90 was at a discount, then he has been super-highly overpaid.  And anyone at 4+ years has been so as well.

The fact of the matter is that almost all players’ contracts at 3+ years can be explained by my model, or something with a tiny bit of discount.

Now, while 1/20 and 5/90 would be equivalent, let’s not forget that all free agent contracts are overpriced.  The 1/20 means that it’s 10MM overpriced.  5/90 means it’s 45MM overpriced.

I don’t want to call Torii’s contract a discount, I see it as a bit of an overpay.  Probably best to ignore what I wrote in #219, think more about it it doesn’t seem to make sense.

I’ll stick by one minor point though, and I’m not arguing that the model doesn’t work.  I just don’t think that Torii Hunter would view 1/20 as equivalent to what he signed, and if another team offered him 1/21, I don’t think he would have considered it for a second, other than as a start to see what kind of multiyear offer he could get.

Which is why I said in 220 that he wouldn’t consider it.

He knows that free agents are being paid twice what they are really worth in terms of revenue generating, which is why he’d go for the 5/90 deal, even if he’s offered 1/30.

Milton Bradley signs with the Rangers for 1 yr. and $5M. That assumes 1.15 WAR. CHONE has him as a +11 hitter, but ZiPS is more pessimistic. Let’s call him +9 runs or .8 wins as a hitter. The fans have him as an above average COF, while PMR lists him as about average. Making him +.2 wins for defense gives him 1 WAA or 2.5 WAR. Less .5 WAR for position gets him to 2.0 WAR, which makes this one heck of a deal.

#223:  Does that include adjustments for playing time?

From 2005-2007, Bradley has averaged 77 games a season.

#223: That calculation assumes 140 games played or so. That probably is high as you point out, but even forecasting him at 100 games still means it’s a good contract. I’d probably put Bradley’s over/under at 120 games anyway.

Yes, great deal by the Rangers.  That’s the sort of thing teams should be doing: taking risks with shorter-term deals on unpopular players.

In thinking about this “discount or not” issue, I had a few thoughts.  One, my article at THT found that long-term contracts were better deals than short-term ones last year, but that may be due to the recent staggering inflation of free agent contracts.  Perhaps Tango is right: there were no discounts in these deals at the time of signing, but inflation turned them around.  Which is one reason teams should sign key individuals to long-term deals: as a hedge against inflation.

Second thought: the difference that Rally and others noted between how much teams are willing to give up for Santana-level talent.  I think this may show that teams really do understand that wins should be valued on an incremental basis (that is, the second win over average is worth more than the first win over average).  As Neyer said in a comment re: the Garza/Young deal, in general whoever got the best player got the best deal.  Same idea.

But, if you believe in Tango’s model, it doesn’t appear that teams and players factor this into contract negotiations and salaries.  As I’ve said before, I think this is partly due to the fact that, if they did, top players like A-Rod and Pujols would break the bank.  Which shows just how out of hand free agent salaries have gotten.

I’ll retract my reasoning about the “overpriced free agents” being the reason that 5/90 is better than 1/20.

I think it’s more of a matter as to what’s the most dollars a player can get, and then won’t go more than a year or two less than that.  If Torii thought that 90MM was the most he can get (implying really 6 years with my model), then the lowest he’d accept is a 4-yr deal.  Anything less than that, in years, means he’d want a premium.

It would be interesting to think what 3yr, 2yr, and 1yr deal Torii would have accepted.

The Bradley deal seems like a great one. What about the Gagne deal?  Other side of the spectrum?

Bradley: 1/5, paying for 1.05 WAR.

Plus hitter, and plus fielder.

(Fielding: Fans, UZR, Dewan in agreement.  Consider him to be an average CF, or a plus corner OF.  Overall, his fielding+position is plus.)

An average player, expected to play only half the time, would be a 1 WAR player.  So, this is really the prognosis here.  Bradley, over the last three years, has played only half the time.  And, it seems that his knee surgery are turning his otherwise healthy body’s plus hitting, plus fielding game, into an overall average player.

There is an enormous upside here of course.  He can easily be a 3 WAR player, playing full time (worth 13.6MM).  He can just as well easily be a part time 0.5 WAR player (2.6MM). 

This has alot of uncertainty, and obviously, the market for high-uncertainty players is low.

***

Gagne: signed for 1/10, paying for 2.2 WAR.

For a closer, a 2.2 WAR implies 72 innings at .655 performance.  If this was 3-5 years ago, fine.  What a truly horrible deal.

Either my salary model is crap for relievers, or MLB teams have stuck their heads up their -ss-s with relievers. 

Out of all the relievers I’ve noted in this thread, have I liked any?  I don’t remember.  I think half were on the border, while the other half were bad.  I don’t think there was one good reliever signing, has there been?

In post 93, I said this about Aaron Rowand:

I wouldn’t be surprised if a team will be signing for both his legacy as a great fielder, and his great hitting in 2007.  aka, he’ll get the Gutsy White Guy contract.  Who gets the Underachieving Black Guy contract?  Because, obviously, you can’t have the converse.

I guess the answer is: Milton Bradley.

Anyone want to offer the winner of the Underachieving White Guy contract?  I don’t think there is one. Eric Gagne should have been the one, and he somehow managed to get the Gutsy White Guy contract!  If Eric Gagne were black, do you think he would have only found a 1/3MM deal on the table?

Am I really just seeing things?  Am I being as racially biased (the other way) as those I’m talking about?

You can’t attribute it all to race, as there is no other player of any race who compares to Bradley in being both injury prone and a headcase. 

I know the umpire was wrong in that situation as well, but the last image anyone saw of Bradley in 2007 was him tearing his ACL while trying to go after an umpire.  Its hard to find a better illustration than that of being an injury prone headcase.

I wasn’t only talking about Bradley.  I also meant Gagne.  Isn’t his lasting image in a Redsox uniform something even more scary than Bradley?

This goes back to last year, when MLB.com rolled out its “Gutsy White Guy” award.  (Did they also have the “Underachieving Black Guy” award too?) I think MLB.com learned from that, because I don’t think they have the Race award this year.

Gagne can have the gutsy white boy award, at least until we see what Rowand gets.

I’d rather have Percival than Gagne, and he only got 8 million - total for 2 years.

"For a closer, a 2.2 WAR implies 72 innings at .655 performance.  If this was 3-5 years ago, fine.  What a truly horrible deal.”

How bad can a one year deal ever be, though?  If the Brewers had signed him to a 4/$36, that would be a “truly horrible” deal, IMO.  Melvin knew that he had to overpaid to get the one year deal that he wanted.  I liek this deal a heck of a lot mroe than the multi year deal Riske got because no matter how big of a mistake the Gagne signing was, it’s over after this year.  And if Gagne has a decent year, the Brewers will get a draft pick when he signs a long term deal somewhere else (what is the cash value of a draft pick, anyway?).

Also, I’m usually the last person to consider this kind of stuff, but there is some value in the Brewers outbidding everyone else for the services of a “big name” player.  That is one of the largest complaints I hear from the casual Brewer fans and a signing like this helps to change the perception that the Brewers will never be able to compete with the big boys.

It sounds like every reliever the Redsox doesn’t want is signed to a 10MM+ deal, and every reliever they do want is signed for 3MM or less.

Gagne has alot of potential obviously.  You don’t pay Wagner and Hoffman money for potential to recapture past glory.

Yes, if you are going to make a mistake, make it a 1 yr mistake.  But, was there actually a team out there ready to give Gagne 9MM?  If there was, then that’s two GMs I’d fire.  I’d be surprised if there was a team out there ready to offer him more than 5MM for a 1yr deal.

Gagne has, what, a 16.7% chance of recapturing his glory?  Hope the Brewers roll a six.

***

As for the “see, we get a free agent name”, I don’t buy it.  Spend 50MM, then let’s talk.  It reminds me when we spent big money on Graeme Lloyd and Toad Irabu.  No way a free agent looks at the Gagne deal and says “hey, let me consider… uh, who did he sign with again?… oh year, brewers”. 

Players are all about one thing: money.  And then, they are all about a second thing: explaining that his family comes first, and he needs to talk it over with them, when really, it’s about the first thing.  And, they are all about a third thing: chance to win a ring, when really, it’s about the first thing. 

You tell a player in his early 30s or younger: “I can guarantee you 150 games, and you expect a 10% chance at making the playoffs, or I can guarantee you that you might get 50 games, for HALF the money, with an 80% chance at the playoffs”, what’s he going to say?  He’s taking the money.  A guy who can sign a 15MM a year deal will not accept less than 14MM for that second scenario.

Players simply don’t leave money on the table for a great chance at a ring, until they are in their late 30s.  And even then, less than half would do so.

"As for the “see, we get a free agent name”, I don’t buy it.”

If that was in response to my earlier comment, I was talking about the perception of the Brewers by its fan base. Whether it’s logical or not, the casual fan is going to look at this signing and think it means the Brewers are finally, “committed to winning”. When you have an organization that hasn’t made the playoffs for 25 years and just had it’s first winning season (83 wins!) in 15 years, overspending by $6 mil may not be the worst thing in the world.  Either way, we aren’t talking about the average organization here.

I’m not trying to convince anyone that this signing makes sense sabermetrically (it doesn’t); only that a GM and/or owner(s) may sometimes have to make other considerations beyond simply production vs. cost.  For a one year deal, it’s much easier to justify doing so, IMO.

Anyone want to offer the winner of the Underachieving White Guy contract?  I don’t think there is one.

Underachieving White Guys:
1.  Michael Barrett (resigned w/ SD, terms unknown)
2.  Brad Wilkerson
3.  Morgan Ensberg
4.  Shawn Green
5.  Marcus Giles

Hallmarks of Underarchieving White Guy:
1.  Strikes out a lot.
2.  Plays for the Padres.

Did any one of those sign for below market value?  Looks more like they keep getting chance after chance.  Might as well add Jeff Weaver to that list.  Him and Pineiro are the Gutsy White Guys.

Giles’ value right now is the league minimum.  Because of his defense, Green is a replacement level player.  Ensberg used to be really good, only a year and a half ago, but nobody seems to want him.

He may be the underrated white guy, a projection on him is still good but seems like everyone in baseball thinks injuries have ruined him, and maybe they are right.

Does Carl Pavano count?

Concur that the Gagne signing is bizarre.  But it may end up looking good.  The translation of his performance in the AL last year to the NL with him being placed in the closer role would show him with good ‘conventional’ stats.  The Brewers may win their division and enjoy a consequential attendance spike. 

The use/valuation of relievers of the last 2 decades mirrors the use/valuation of leadoff men in the 60s.  A generation of mostly smart people (GMs) cannot escape stupid industry norms of the time.

The Underachieving White Guy contract is the opposite of the Gutsy White Guy contract.

It is a contract signed at below market value for a player who performs fairly well, but is given a huge discount for being something… usually Black, followed by something-else.  A White guy with that same something-else is not signed at such a discount.  Well, that’s what it looks like to me anyway. 

So, Joel Pineiro qualifies as signing a Gutsy White Guy contract.  Gagne, too.  The question I’m asking is: which free agent White Guy has signed a contract at a heavy discount?  Anyone?

Let’s not even limit it to this year.  Who is the last White guy to do so?

I looked through last year’s thread.  The best I could come up with are some arbitration-eligible guys (Utley, McCann, Mauer) and then of course Tim Wakefield.  I’m not sure he counts as white though, aren’t knuckleballers a species unto themselves?

Red Sox also got a discount on Schilling, maybe Timlin.  That annoys me.  A team that has a payroll well more than every team but one should not also be getting discounts.  That makes a 150 million payroll effectively 165 or more.

Yeah, I wouldn’t go with arb-eligible players.

I was thinking of Schilling, but he did get some bonuses.  It seemed like a reasonable deal.  David Cone also gave a discount near the end.  But, these guys are pushing 40.  They got a discount for being an old pitcher.

Wakefield though qualifies.  I am shocked that the MLBPA would allow the reserve clause to be actually written into an agreement after fighting so hard to remove it.  Wakefield has allowed a clause that the Redsox can renew his contract, year-to-year, at 4MM (no escalators).  And in return, the Redsox… uh… the Redsox.  Nothing.  They don’t grant him any concession for including the clause.  No buyout for number of years extended, no nothing.

Some people actually argue it’s a good thing for Wakefield to have this clause!  Wakefield could have instead filed for free agency year after year, and simply signed a one year deal from the Redsox (even at 4MM).  He certainly would not have signed for less, since the Sox would simply let him go if they didn’t think he’d be worth it.  He gets no benefit here.  None.

The Underachieving (or Underappreciated) White Guy contract was signed with a Slavery clause.  How appropriate.

Good job Rally.

Tejada: 2/26, paying for 3.1 WAR.

His Marcel is a .356 wOBA on 555PA (implies just under 1 win in hitting), includes aging.

Fans have killed his fielding since 2005 (dropping him 9 runs each year!).  They’ve basically see him as +1 win relative to other SS in 05, to even in 06, to -1 win in 07.  Quite a collapse.  Dewan and UZR don’t show this tailoff, but that’s to be expected with sample data.  They see him at just a shade below league average.  Let’s call him at a shade better than -1 win in fielding, including aging.

He gets his +0.5 for doing that fielding as a SS.  (Won’t matter if he’ll be a 3B, since his fielding+position remains fairly static.)

So, he’s a +0.5 WAA player in full-time play.  Add in 2.25 WAR, and that’s 2.75 WAR in full-time play.  He did that in the AL, so that’s +3.00.  Marcel forecasts 80% of games played, so he’s +2.4 WAR.

His 2yr contract is really worth 20MM.  So, the Orioles managed to sell an asset with a 26MM mortgage on it, but worth only 20MM.  Fantastic if this were only a salary dump.  On top of that, they get below-market assets on top of that (I don’t know how much they are worth, but almost certainly they are a plus).

Looks to me that the Orioles stole some 10MM from the Astros’ pockets.

Orioles did Ok, but I think you’re being too hard on Tejada.  80% playing time is 129 games.  I’ll gladly take the over on that since before his one injury he played 162 every year.

I know you can’t forecast a guy to play 162, but I would be comfortable with Tejada at 145-150 games.

As a hitter B-ref has him at +25, +26, and +6 the last three years.  I don’t think this is overly optimistic:

+18 hitting
+0 position/defense
-5 age
+35 over replacment/162

In 90% playing time that’s 3.1 WAR

I have Tejada as a win in hitting.  His UZR is -2, -4, and -6 the last 3 years but +10 in 04.  I’ll give him -.5 wins for fielding and baserunning.  Add in .5 win (I usually use 1 win for SS) for position, and he is 1 WAA.  That is around 3 WAR.  This is all per 150.

His offensive lwts last 3 years are: 30, 23. 16. 3, from 04 to 07.

You may very well be right in this particular instance.  Marcel isn’t particular.

If I can remember to bring in my database, I’ll look for quality players like Tejada: at least 500 PA in season x, after having seasons of at least 100 more PA in seasons x-1, x-2, x-3, and be an above average player.  Unless someone out there wants to do it.

The basic Marcel rule (50% of PA in season x, 10% in x-1, plus 200 PA) may not apply to someone in Tejada’s circumstance.

MGL is saying 1 WAA (per 150), Rally is saying +1.3 (per 162), and I said 0.5 (per 162).  So, the difference is mostly tied-in to his fielding.  I’m relying heavily on the Fans’ assessment.  We’ve got a 0.5 win difference here.

The other disagreement is on the playing time.  Marcel kills him (555 PA, which is 79% of 700 PA or 162 full games).  Giving him 80% or 90% means 0.3 wins.

Adding up, and it’s 0.8 win difference for things that we are really just a bit different on.  Incredible how things add up.

So how much more is Tejada worth than Adam Everett, whom the Astros will presumably be letting go?

Maybe the biggest thing I’ve learned from this thread, and last year’s, is how sensitive salary is to relatively small differences of opinion in a player’s true talent.  Shows how much money is being thrown around, and how leveraged the free agent market is.

Aaron Rowand, 5yrs/$60M, paying for

Offense: Marcel (.347 wOBA, +4)-- CHONE (.345, +4)
Defense: PMR (0)-- Dewan (+5)-- UZR (+15)-- Fans (0)
Other: Position (+7)-- Aging (-5)-- League (-2)

So, that gives me a total of 10 RAA (4+6+0) or 1 WAA. That’s about 3.2 WAR for 155 games or 2.8 WAR for 136 games. 2.8 WAR yields about $45.5 million for five years, so this is an overpay.

Why 136 games?  I was thinking 5/55 but I’m pretty sure the whole difference is that you subtracted 20 games.

Great point, studes.  Overpaying one player to fill a hole can be forgiven, but a team that consistently mis-values players gets killed by the free agent game.

According to UZR, Everett’s in the +35 to +40 runs per 150 games range, with 2006 sitting at +51.  Assuming his bat is replacement level (is it?), he’s better than Tejada by a full win.  If the Astros non-tender him, make him a utility infielder, or trade him for next to nothing, they’ve made a huge mistake.  Given their current pieces, Tejada should play third, Everett short, and let Wigginton/Matsui piece together an offense/defense platoon at second.

254/
Marcel projects him at 586 PA which is good for 140-142 games for 2007. But it’s a five year deal, so I felt the need to use a lower total to represent the average number of games played over the life of the contract. I’m fairly comfortable with that figure. At 145 games per season, the “fair” market value is $51.2 million (3 WAR).

Rowand: 5/60, paying for 3.7 WAR.

Phil did a great job, except on the aging.  It’s already included in the offense part for Marcel, and I give CF +0.5 wins. 

In post 93, I said 4/36.  A huge overpay I think.

A guy like Everett (no hit, great defense) is going to be WAY underpaid.  Teams cannot get their arms around what great defense is worth because there are no traditional stats that they can easily use.  The equivalent of a great defensive player like Everett is around 35-40 HR and 100-120 RBI.  Now teams can dish out 15-20 mil a year for a player with those offensive stats.  They ain’t going to dish out half that much for a “slick fielder.” Not to mention the fact that while Everett is the greatest defensive player of the 21st century, bar none, he is NOT considered as such (for whatever reason), evidenced by his lack of GG awards.

I have Everett at almost 3 WAR or a 15 mil a year player.  Who is going to pay that kind of money for this guy?  He’ll be lucky to get 1/3 of that.

I love Rowand.  With almost a win on defense, a half win on offense and baserunning, and .3 wins for position, I have him as 1.9 wins AA or 3.8 WAR or worth 19 mil!

His last 4 years offense:

18
-7
-6
25

His last 4 years UZR:\
13
30
-9
13

With a +1 run arm per year.

Tango, I see $65M at 5 years, 3.5 WAR.  How is $60M paying for 3.7 WAR?  Am I misreading something?

I’m not as good with the math as most here, but I’m with MGL on Rowand. As far as I can tell, he’s a better overall player than Torii Hunter. I’m surprised the fans gave Rowand a 0 this year. I watched nearly every Phillies game (admittedly I did that BECAUSE of Rowand - I’m not a Phillies fan), and Rowand has maybe lost half a step over his best defensive years but that’s it.

As for his offense: Is it possible that 2005 was an aberration? He had a great 2004 and a great 2007. He suffered two pretty serious injuries in 2006. I know that injuries should be factored into projections, but I’m wondering if his down year in 2006 offensively could be attributed to the injury more than anything else. Basically, I think Rowand’s true offensive talent is probably a lot closer to 04/07 than 05/06.

Relative to other center fielder contracts given out over the past year, I think this is a bargain. Eric Byrnes got 3/30, and I think he’s pretty mediocre (at best) at just about everything except stealing bases and diving after a misplayed fly ball.

Sky,

If Everett was a league average defensive shortstop, his bat would put him below replacement level.

By the CHONE projections Everett is a -25 hitter, Tejada a +14.  But on defense I’ve got Tejada as a zero and Everett a +31.

Astros didn’t improve themselves much, if at all, they just traded defense for offense.

Studes: darnnit.  Yes, I updated my salary chart on my computer to include the “discount” we were talking about, and left it at 10%.  So, a 5/65 is a 3.44 WAR (including the min salary of 0.40MM, which is not on the website).  So, the overpay is not that bad.

As for his fielding, Phillie fans LOVED Shane Victorino and were almost as high on Michael Bourn.  So, they have their frame of reference.  Rowand didn’t measure up to them.  (They also loved Rollins.)

Same thing with Tejada.  That should read: 2/26 paying for 3.0 WAR.

JD, don’t ever think of a good or bad year as an “aberration” (and then try and ignore or discount it).  Every year’s performance contributes to a player’s projection - period.

Rodriguez finalized his 10/275 deal. According to the details in the article linked on my name, his salary breaks down like this (assuming 10% inflation and constant $400k league minimum):

Year Salary WAR
2008 $29 6.5
2009 $33 6.7
2010 $33 6.1
2011 $32 5.4
2012 $30 4.6
2013 $29 4.0
2014 $28 3.5
2015 $21 2.4
2016 $20 2.1
2017 $20 1.9

That salary progression makes me believe they’ve been using my salary chart!

That’s a total of 43.2 WAR.  You can also get a 43.2 WAR by starting at 6.57 and dropping by 0.5 every year.  THAT would imply a 10/280 deal.

I’m presuming that the “achievement” money was the Ranger subsidy.  That is, in order to save face on losing the Ranger subsidy, they did this backend thing.  I think we need to include a discounted portion of this money, since ARod’s got a certain probability of achieving each one.  My guess is that the actuarial value of the 30MM bonus money is worth 21-22MM dollars (i.e., Rangers money).  Adding that in, and he’s being paid for almost 7 WAR.

I always want to point out that this presumes there is no discount given by ARod for signing a long-term deal.  If there was, that would imply he’s really an 8 WAR player.  I really don’t believe this is how they approach it.  I think it’s what I said earlier: there is no discount for the correct length, and there’s a premium for short-deals.

MGL,

I didn’t mean to imply that I thought we should ignore a season, but rather maybe it shouldn’t be given as much weight as normal. I just wonder, in general, how often 3 years ago is more significant than 2 years ago (or 2 years more than 1, etc.) when projecting a player. It seems to me that there have to be some players who have a season that should be given less weight than in an average situation.

there is no discount for the correct length, and there’s a premium for short-deals.

???  Aren’t they the same thing?

I think of “Happy Hour” as a discount, and the rest of the day as regular-priced.  That’s where I’m coming from here.  Most players are getting lengths of deals that are regular-priced.  Premiums are applied only for short-term deals.

Same applies in the consulting world.  You pay a premium for a 3-month contract, but the regular rate for 12-months or longer.

Tango/269—excellent happy hour metaphor.  I’m using that one going forward.

Rally/261—your offensive numbers are relative to average, correct?  So if Everett is a league-average shortstop, he’s only about five runs worse than replacement level overall.  Gotcha.

Anyone heard any news on how Everett’s recovering from his broken leg?  Will he be the same guy for the Twins?

JD, well, that is what I said, right, ignore or discount?

Discount="not given as much weight as normal.” Right?

Of course that is right that not every player should have the same projection algorithm.  The problem is that we don’t know for which players we should use something different, so we just use the “average algorithm” for everyone.  That is true for lots of things, like aging curves.

Now, if someone does research and finds that certain subsets of players (with certain characteristics) have different algorithms that seem to work better, than that’s fine.  For example, catchers probably have significantly different aging curves than other positions, fast and slow players have different aging curves, etc.

It might be that if we test the weighting for past seasons, that if we “flag” a season as an “injury season” being careful not to base that assumption in any way on a player’s stats (otherwise we will tend to ignore seasons where a player was injured but had good stats and erroneously flag seasons with bad stats as injured), it might be that we have to “adjust” the weight of that injury season.  Who knows.

But to just isolate a grossly anomolous season (without having any other information) and suggest that we should give it more or less weight, we know that is wrong.  Tango, I, and others have done extensive studies on “banner” (good and bad) years.  In general (not considering age, experience, etc.), the normal weights (around 5/4/3.2/2.6) still work best.

I think of “Happy Hour” as a discount, and the rest of the day as regular-priced.  That’s where I’m coming from here.  Most players are getting lengths of deals that are regular-priced.  Premiums are applied only for short-term deals.

Still confused.  In the end, it’s the same impact, right?  Higher salaries for shorter contracts.

My defensive projections have Everett at +31, which is just an insanely good projection, but he’s been an insanely good defender.

He’s getting a 1 year, 2.8 million dollar deal with the Twins, or about +6 runs over replacement.  If we can agree on his -25 offense (which is pretty much his career average) then Minnesota is expecting a +5 to +10 defender, and nobody else thinks he’s better than that.

Either they don’t agree with us on the quality and value of Everett’s defense, or the teams have serious doubts as to him continuing to be so great after the broken leg.

One, they have no idea how to quantify defense AND they simply refuse to pay a lot of money for a no-hit, great-fielding player, especially a SS.  Not to mention the fact that they do not differentialte between guys like Everett and Rollins.  Everett is simply thought of as a slick-fielding SS, but not one of the best ever, which he is.

Also, teams’ assessemnt of value is NOT linear.  By that, I mean that if a fielder were a true +40 defender (whether that is possible or not), there is NO team that will pay him a lot of money if he hits at replacement level, even though he is obviously worth a fortune.  Teams do NOT add up offense and defense - not that they would know what to add anyway.  They STILL think of offense as the principal basis of player evaluation and defense is sort of an afterthought.  They also “back into” linear evaluation of offense because they can plainly see a continuum of numbers like BA, OBP, OPS, RBI, HR, SLG, etc.  Not so for defense.  Players tend to be grouped into one of 3 categories on defense -average, below-average, and slick.  Fantastic defenders (like Everett) and terrible ones (like Griffey and Manny) will ALWAYS be under or overvalued, again, by mainsream teams. 

You cannot SEE great defense in the sense of there are no garbage numbers.  In fact, you cannot see great range in the sense of any numbers at all.  Great offense, howewever, sticks out like a sore thumb.  Unfortunately terrible offense, like with Everrett, also sticks out like a sore thumb and no mainstream team in baseball thinks that great defense can possibly make up or even surpass (which in Everett’s case, it does) stinky defense.

There is no way that any of the mainstream teams accurately assess Everett’s worth.  No way. Not even close.

Fans think that this signing is “questionable.” Even somewhat knowledgable fans.

Not only is it not questionable, it is the steal of the century, literally.

Funny, I have him at exactly +31 (like Rally) in UZR projection (per 150).  That is with regression and age adjustments!  He may be the best defensive SS of all time. 

I have him at -32 in hitting, which is replacement level even for a SS.  So basically he is league average.  Add in .5 to 1 win for being a SS, and he is a 2.5 to 3 WAR player.  For 3mm dollars!  OMG!

If I told a mainstream audience that he was worth almost as much as Jeter (which he is - I have Jeter as .5 win better for 08), how long would it take for them to stop laughing?

Note: And that is with a fairly conservative UZR projection for Jeter, -13 per 150.

UZR per 150

04 -1
05 -11
06 -15
07 -31

Everett UZR per 150

04 +18
05 +28
06 +49
07 +16*

* Limited # of games (60 or so).

Since 1993, I have Everett and Sanchez as the two best fielding SS in baseball.

***

I evaluated the Everett signing last year:
http://www.insidethebook.com/ee/index.php/site/comments/sabermetric_moves_of_the_off_season/#108

I called him a 2.5 WAR player.  (I don’t know why I said 1/2.8 paying for 2.3 WAR as a 4+ player in that post.  I must have meant he was 3+ at the time.)

He’s at 5+ years of service, meaning he will be a free agent after 2008, and deserves 80% of free agent value for 2008.

You’d think that a team like the Astros seeing Everett on a day/day basis would start to appreciate this unique talent. I guess not, or they feel his leg injury will be too difficult to come back from. I imagine however, that conventional wisdom ruled the day, it certainly does at my work-place

Ironically the one team that probably should carry a no stick/great field player is the Yankees, move Jeter to left, put Everett at short, and with their offence a better team.

MGL given your unique history, what do you think of Eckstein to the Jays?

I’ve got Eckstein -9 hitting, +3 fielding.  Also -2 last year on DP runs, a new thing I’m doing from retrosheet data but don’t have multiyear or projections on that yet, but I’ll use it anyway for now.  Add +5 for SS and +20 for replacement, and he’s a +17.  I like a higher number position adjustment for SS (+9).  Either way Jays get a bargain since they are only paying for one win.

Well, I don’t have any inside information concerning Eckstien.

He probably lost whatever above-average SS skills he had and I actually recommended to STL that if they should re-sign him that they consider platooning him at second with Kennedy, although that might be too harsh, moving him to second that is.

He is what he is of course.  I have him as a little worse than an average SS, which is not too bad.  Probably worth at least 8 mil.  Pretty good bargain.

I have said for a long time that corner OF’ers and first baseman tend to get overpaid.  If that is the case, then the other positions will be underpaid, especially SS and CF (I am not sure about C).  Enter Eckstein and Everett.

"Well, I don’t have any inside information concerning Eckstien.”

I was hoping to hear about the time you walked into the dressing room and saw Pujols rubbing flax seed oil all over Eckstein, but I guess these things are better left behind the locker room door.

MGL,

Thanks for the explanation. It’s in my nature to question everything, even the sabermetric numbers that I trust to be right (or, at least, better than the alternative).

So my next question is how much work has been done regarding injured players and projecting them going forward? This might be a bad way to look at things, but if you went back to pre-2007 and looked at Rowand’s last 3 years (04-06), would anybody have projected the season he had? So does this mean he played way above what he should have or would too much weight have been given to the most recent injury plagued season? I guess that question is sort of rhetorical. Plus, I might be looking at things wrong by working backwards like that.

I’m comfortable accepting the “average situation” when it comes to, well, everything (you might be surprised how many people aren’t ok with that). I just haven’t seen a lot on certain not-so-average/typical situations.

By the way, I should add that I got The Book about a year ago. It was a thoroughly enjoyable and insightful read. I appreciate all the work the three of you have done.

JD, thanks for the kind words.  It is certainly OK to “play around” with various weighting algorithms for various scenarios and see which ones work best.  Or run some regressions on the data and see what kind of equations come out of it.

There are (at least) two ways to come up with various weighting algorithms which might be appropriate for various groupings of players.  One, isolate the grouping first, such as players over the age of 30 (or under), and then run a regression, or play with various weightings and see which ones work best (provide the most accurate projection).  Two would be to run a regression including the “grouping” (e.g. in this case over or under 30) as a regular or a dummy variable.

You can do the same thing for players with seasons in which they are injured.  For example, set up a regression with a yes/no dummy variable for an “injured” season.  Or play with increased or decreased weightings for injured seasons. Etc.

The most important thing that mgl said was this:

It might be that if we test the weighting for past seasons, that if we “flag” a season as an “injury season” being careful not to base that assumption in any way on a player’s stats (otherwise we will tend to ignore seasons where a player was injured but had good stats and erroneously flag seasons with bad stats as injured), it might be that we have to “adjust” the weight of that injury season.  Who knows.

He’s talking about selective sampling here.  You can’t use knowledge of the data you are looking at, to establish some other knowledge, since you will only be focus on some of the samples to begin with.

You always hear the b.s. about “I saw the ball as big as a balloon… that’s why I hit 3 HR today”.  Once, just once, I want to hear a ballplayer say “I saw the ball as big as a balloon… and yet I managed to strike out 5 times.  I’m a horrible ballplayer”.

It simply won’t happen.  If you hear the stories, you’d think everyone makes out great in Vegas.  It’s the same thing here.  You’ve gotta account for all sides of the ledger.  You can’t just focus on some non-random sample, and then treat that as a representative sample of the population.

This is a “big picture” type question. Baseball is drowning in revenue. Is there really anything wrong with overpaying for a free agent?  For instance, the Reds had money to spend. They were going to spend it on something. They had a need for bullpen help. They bought a very good arm (the best one available). They weren’t going to get Cordero for much less than what they paid. I think most would agree the addition of Cordero makes them better. Where is the problem?

If you have 100,000$ in your bank account, and you really need a car, would you give that to your Honda dealer for a Civic?

Sure, it puts you in a better position today, but what about tomorrow? 

The “problem” is the imbalance.  And, the “problem” is that teams don’t need to jump over themselves to throw money down the toilet.  If they applied any kind of rationality whatsoever, they wouldn’t need to spend so much money.

At the same time, teams don’t really lose anything for overspending on players.  Yes, it’s not cash as well-spent as it could be, but I’m not aware that any of this has an impact on franchise valuations, which is all an owner cares about financially, in the end.

In fact, arguably you could say that owners hurt their franchise valuations when they stay out of the free agent hunt because it undermines the profile of the franchise.

It is a complicated question/issue, Mark.  For one thing, it depends on your goal (utility I guess the economists call it).  If your goal, as with most businesses, is to make as much money as possible given a certain capital investment (ROI), then you simply hire all players whose salary is less than what you expect your increase in revenue is for that player versus not paying anyone at all or versus a replacement player.  That is why we usually use WAR as our measure of value for a player.

Clearly that amount (how much a player is worth in extra revenue) varies among teams with a lot of things and the variability is probably enormous.

Now, even in that model, you have to always weigh the alternatives.  For example, if you can hire a FA who is worth 2 WAR for 12 mil and that will increase your revenue by 13 mil, it is a good hire.  But if there is another FA or two who will also give you that same 2 WAR, and they are available at 10 mil, then it was a bad hire.

You also have to weigh FA hires against developing in house talent, trades, you have to weigh short term against long term (most business look toward maximizing long term - maybe 10 year - profits, right?).

Now the other model is to simply field as good a team as possible or a team that has the best chance of making the playoffs and winning a WS, given a set of constructs, such as a certain salary budget.  Many teams seem to operate from this model, or at least some combination of this and the previous (bottom line only) model.

The issues and alternatives with this model are similar to the bottom line model, the main difference being that the team does not necessary have to worry about paying a player (or otherwise investing money) more than that player will provide in marginal revenue.  If I have 60 mil to spend (say I am the GM) and I am told to field the best team possible (again, short and long-term considerations complicate the matter), then I simply find the most marginal win value I can with that 60 mil.  That is by far and away the easiest model to work with and if I were a GM I would be thrilled with that model.  Of course, all the other issues apply, making it not so easy, such as balancing long and short term goals, rebuilding or not rebuilding, how much money of that 60 mil to put into player development (although if that is a separate budget it makes it a lot easier), etc.

So, yes, within those two models, or any other models, an optimal strategy can certainly include overpaying for players.  That is certainly the case with model #1.  If you are the Yankees and it is economically in your best interest to field a ridiculously good team, then you probably are going to to be forced to overpay a lot for some players, especially in years where you are starting out with a weak team (and you have to acquire A LOT of talent, i.e. marginal wins).

What we are mainly talking about here, though is how much money a certain number of wins should be able to be gotten for (sorry for the awkward sentence).  IOW, if a 2 WAR player is overpriced, a smart team can probably wait for another player.  Sure, if a team SHOULD (via either model) acquire 2 WAR and the only players that are available are overpriced, then it might not be a mistake to overpay for those 2 WAR.  That is not usually the case though.

An overwhelming majority of times that a player is paid more or less than he is worth based on our model here (which requires estimating a player’s projected WAR for one or more years and assigning a dollar value to one WARm usually in the 4-5 mil per year range), it is because these players are over or under valued by teams who do NOT know how to evaluate some or all players for various reasons.

Houston did not acquire Tejeda for 11 mil a year and release Everett who would make 3 mil a year because they HAD TO.  They did that because they have no idea how to evaluate players (or at least a player like Everett).  (Of course we could be completely wrong about Everett’s defense at this point in his career - not so likely though.) If God came down and told them that Everett was worth around as much as Tejada, then they would have tendered Everett.  Or more likely, they would have told God to, “Get his head back in his spreadsheet....”

There’s no reason for it to necessarily impact franchise valuations.  The value of an asset is the sum of all its discounted future earnings.  Unless you are saddled with 12-yr contracts, the value of a franchise would only take a short-term hit.

However, you are losing operating income, since you are spending more than you need to.  That is, throwing away 10MM to Gagne, when 6MM would have been sufficient, etc.  So, that’s 4MM less for the current owner, with no impact at all to the franchise valuation.

If Gagne is making too much, someone must be making too little.  Unless you are suggesting that it’s a necessary evil for Gagne to make too much, to further the cause of the franchise valuation.

I think that’s called a bubble.

If so, it’s a bubble that’s been in effect for decades.  The lesson I take in the never-ending increase in franchise valuations is that it’s better to have your cash in player contracts rather than cash, particularly when governments and/or your major league cronies will bail you out.  As long as you don’t go negative, you’re golden.

I don’t think we’ve proved a causation between franchise valuation and overspending on payroll.  Overall, the average team has risen by 10% in value, 10% in revenue, and 10% in expenses. 

At the same time, out of pocket money is still in the red, which is balanced against the cache of owning an MLB team. 

I’d be more interested in seeing a team with positive operating cash flow compared to a team in the red, and seeing how the change in valuation might be different.

I generally agree, Tango.  We/I haven’t proven anything.  But the scenario I’m painting is one that I believe could drive certain economic behavior.  To me, it’s as plausible as the standard/solid economic behavior that you and MGL are describing, given the behavior we see among major league teams.

I’m responding to Mark’s original question, essentially stating that overspending may not be a problem for certain teams and owners.

Question.  When you say that “out of pocket money is still in the red,” do you mean you believe that ML teams operate at a loss?

I know it’s a little difficult to analyze a trade (especially one involving minor leaguers) about 10 minutes after it happened, but as an A’s fan I’m loving the Haren trade. It doesn’t make the major league team better in 2008, but in the long run it looks fantastic.

Chris Carter looks like a stud, Aaron Cunningham makes me giddy, and Brett Anderson just needs to overcome that whole high school pitchers and exploding limbs problem. Greg Smith doesn’t look half bad either.

Thoughts?

Gonzalez is the most interesting player. Carter and Cunningham probably can be useful and Anderson has a shot at becoming Blanton lite. The other guys are guys.

I guess I’m a little surprised that Oakland didn’t get something cooler....

There are better people to estimate this but here’s my thumb in the air. Haren might be worth something like 18-20 WAR in the NL over the next 3 yrs.  I don’t really see the As return blowing that out of the water. That said, if Gonzalez hits his ceiling, his 6 yrs beats Haren’s three.

Dan Haren gets paid 16 million over the next 3 years.  He’s an excellent pitcher, his market value is probably something like 18 million per year, so the A’s need to get 40 million in surplus value to make it work.  Did they?

Probably. Figuring Haren’s value is the easy part.  One young player with no service time who is an average player is worth, what, 20-30 million in surplus value?  Tango put a number out there

For the youngsters, especially for those far away from the majors, they are not league average right now, maybe not even replacement level.  Not because they are bad, simply because they aren’t ready yet.

Pulling numbers out of my ass, a good, non-elite prospect in A ball might be 2 wins below replacement.  Apply standard aging progression and maybe he gets to replacement level at his peak.  So he’s worth nothing, it wouldn’t matter if you got one or 100 prospects like that.

But wait, not all prospects develop equally.  Maybe one gets to league average, one becomes a decent bench guy, and another becomes a +1 (+3WAR) player in a few years.  To balance it out, maybe the other 3 don’t even make the majors, but you’ve still got a pretty good haul of talent in your trade.

I can see the value of percentile forecasts here.  I doesn’t do much good to say Carlos Gonzalez projects as a -19 outfielder (or -4 vs replacement level).  To properly value him, we’d need to estimate what chance he has of being a +1 WAR, +2, etc. within 2-3 years.

I don’t think PECOTA even does that - it gives an estimate of his true talent and what he’d do if he gets lucky or unlucky.  At least that’s how I read it - I don’t think the 90% forecast is telling us there’s a 10% chance his true talent is at that level. I know for sure CHONE doesn’t even attempt something like that.

There really is no difference between projecting the chance that a certain type of player gets lucky or attains a certain (good) true talent level for X amount of years. You simply project the chance that a player has a certain total WAR value for his first 6 years in the majors (or whatever), whether that be by luck, true talent, a combination, hook, crook, or whatever.  That is fairly easily done I guess.  Just look at history and identify similar propects and see how many produce X value (in WAR) in the majors (in their slave years), Y value, Z value, etc.  Maybe not so easy for lack of data, but the methodology is fairly simple.  I suppose you want to try and match up A prospects with A propsects, B with B, etc., in terms of finding similar players.  Has anyone ever done that?  I am not aware of any research along those lines.

And it is not that “statistical analysis is NOT immune to common sense (which implies that if common sense contradicts statistical analysis, then common sense prevails).” That simply isn’t true.  BAD statistical analysis or BAD conclusory statements after good analysis is NOT immune to common sense.

Good statistical analysis is what it is.  It recognizes two important things.  One, that almost every statistical problem is sports analysis is Bayesian.  For example, there is a huge difference between a “finding” that there is a statistically significant “split” for odd and even days for pitchers and a similar platoon “split” for pitchers.  It is like flipping a coin you pull out of your pocket and getting 8 heads and 2 tails versus picking out a coin from a magician’s or con artist’s pocket and getting the same result.  Bayesian, Bayesian, Bayesian.  The kind of regression toward the mean that we use in projections is really Bayesian probability in disguise.

Two, a good statistician always recognizes that there is always uncertainty in his results or his conclusions when working with sample data.  And if he is writing to a lay audience he should articulate that in some way shape or form.

I always say (at least I should) that there is “a suggestion” or “no suggestion” or “an X percent chance” or “a likelihood,” etc., etc.  With that kind of language statistical analysis should never be at odds with commen sense.  Another way of explaining or defining Bayesian analysis in these kinds of baseball and other sports problems is that it is the melding of uncertainty and commom sense.

In fact, it is usually the other way around and statements like that of James really annoy me.  Proper statistical analysis enables us to discern which things we THINK are common sense really are or are not.  Statistical analysis is NOT supposed to confirm that which we think we already know.  In some sense, quite the opposite.  James himself used to say that whenever he heard something that was an accepted truth, a truism, or conventional wisdom, he would always ask himself, “Is that really true?” And, “How can I really discern whether it is or isn’t?”

The Victor/mgl discussion will be (has been) moved to the BTN thread here:

http://www.insidethebook.com/ee/index.php/site/comments/sabrs_by_the_numbers/

Jenkins: 2/13, paying for 1.6 WAR.

Marcel has him as a league average hitter, with 70% playing time. 

Fans love him in LF, as they would for any white guy who runs like crazy.  Dewan really likes him.  UZR had a love affair with him for the longest while.  Let’s call him at least +0.5 wins as of today, maybe a bit higher.

Corner OF get a -0.5 adjustment, and aging has already been included above.

Overall, you’ve got yourself a league average player, he gets his +2.25 wins above replacement per 162, he loses 0.25 for compiling his stats in the inferior league (gives us a 2 WAR per 162).  Take away 30% for playing time issues, and he’s a 1.4 WAR.  If you want to call him a +1.0 win defender, he works out to a 1.75 WAR player.

Bingo.

Dontrelle Willis: 3/29, paying for 2.4 WAR.

Presumes no aging, cause he’s young (i.e., his upward growth cancels out his potential for injury… I really ought to do aging for guys like him, similar as I did for Buehrle/Santana, to show whether my supposition here is true).

He’s got 4+ years of service, implying 60% of free agent value in 2008, 80% in 2009, and free agent dollars in 2010.

Marcel’s got Willis as a .500 pitcher, with 20.6 full games (185 IP).  Replacement level is .380 (.370 in the AL, .390 in the NL).

WAR = (.500-.390)*20.6 = 2.3

Bingo.

***

Are you guys bored or fascinated by this?  I can’t decide where I am here.

I would love to get an insider’s report about the Willis negotiations, and how many countless hours and days it took to come up with 3/29, when it took me all of 3 minutes.

It’s almost like some big joke.

The last Marcels I saw was for Willis as a Marlin. Does the .500 take into account the league change?

I was thinking 3 yr/$24M was more in line with Willis in the AL.

I have Willis as .2 runs worse than average in the NL.  I have Jenkins as 1.5 WAR before positional adjustment, so 1 WAR after.  I no longer like his defense.  His last very good UZR year was 04.  He is going to be 34 years old!  No way he is 1 win in defense and probably not .5 wins.  I have him projected as average defensively.  I think they are both being overpaid a little.

The Marcels don’t care about parks or leagues.  Even though Marcel should forecast him to be a .480 pitcher now, it won’t.  In effect, Marcel is purposefully giving an edge to the other smart forecasters.  And still, Marcel will do just fine, as he does every year.  Some of the other smart forecasters are actually too smart for their own good.

The salary though won’t change.  Willis is a .500 in the NL and .480 in the AL, making him a .490, which is +.110 wins above replacement, which is what I have in post 300.

A 3/24 would imply a 2.0 WAR pitcher, meaning that he’d be +.100 wins above replacement.  Certainly within the range of uncertainty.  I can accept any forecast that is +/-15% from my Blink estimate.  In Dontrelle’s Blink estimate of 2.3, that means being between 2.0 and 2.6 WAR is reasonable, which implies 2/25 to 2/32.  There’s also the issue of aging for such a young pitcher, and whether I should have used +0.25 or -0.25 or flat.  That would make our new range 3/21 to 3/35.

Tango,

Just a question concerning your Jenkins post (299). You say he should be penalized for compiling his stats in an inferior league, and that makes sense. But if his new team is in that same league should he still be penalized? I don’t understand how him being worse than his stats “matters” since it shouldn’t actually show in his future performance because he’s still playing in that same inferior league.

Are you guys bored or fascinated by this?  I can’t decide where I am here.

Fascinated.  It’s really interesting to see where the market “hits” and “misses.” The disparity with respect to releivers in particular continues to be amazing to me. 

Starter signings like Carlos Silva are also pretty interesting.  I wonder if issues of supply also come into play here.  After all, there are only so many spots on the roster, so it’s not the case that you can always spend in other areas..
-j

Silva 4/48 paying for 3.2WAR

Marcel says hes a .490 pitcher and will pitch 177 innings.
WAR = (.490-.390)*19.67=1.97
So, if the reported deal is true, gross overpay.

Reported here.

Tell me if I messed that up.

JD: if he stays in the same league, his stats say the same.  But, his replacement level becomes 2 wins per 162 G in the inferior league, rather than 2.5 wins in the better league.

If he is traded, his stats change by 0.5 wins per 162 G, but he gets the other replacement level.  So, it’s the same thing.

As an example, say that Jenkins is an NL player, league-average.  He’s a 2 WAR player per 162 G.

If he’s traded to an AL team, he’s now a -0.5 wins relative to an AL team.  But, with repl level at 2.5 below AL average, he becomes a 2 WAR player still.

So, it’s a matter of the perspective in trying to understand what we’re doing.

For Marcel, the forecasts will be definitely wrong.  But, Marcel accepts this inferiority at the price of simplicity.

Silva is an AL pitcher, so his repl level is .370, not .390.

***

Silva: 4/48, paying for 3.1 WAR.

(Note: my website doesn’t show the 400K min salary per year.  My bad.)

Love his 2005 (71K , SEVEN walks), not his 2006. 

3.1 WAR, presuming 20 full games, implies +.155 wins above replacement.  Since that’s .370, that makes him a .525 pitcher.  That’s pretty much the most optimistic you can fairly say about him, given that 2006.

This is a great candidate for Pitch F/x, presuming you had multi-year data.

Marcel makes him a 4/34 pitcher.  I guess his 2006 performance was severely discounted.

I have him (Silva) as pretty bad.  .38 runs per 9 worse than average, or around .8 better than replacement.  For 180 IP, that is around 1.6 WAR or 1.8 for being in the AL.  Gross overpay.

You may be fascinated or bored by the ones that got nailed, but there are going to be plenty of over and underpays to go along with the “nailed its"…

Just to add to what Tango said about the league thing. The average player in the NL gets paid a lot less than in the AL, as it should be.

Which is another, simple, and perhaps the best, way to compare quality between the two leagues.

Prior: 1/1, paying for 0.14 WAR.

(Also incentives to bring it as high as 3MM.)

Basically, there are 4 Mark Priors that you are looking at:
.600 pitcher, 24 full games (WAR=5.0)
.500 pitcher, 16 full games (WAR=1.8)
.400 pitcher, 8 full games (WAR=0.1)
.300 pitcher, 0 games (WAR=0.0)

If we make the odds 1%, 4%, 18%, 77%, that gives us a weighted WAR of 0.14 wins.

I don’t understand how you can have incentive clauses that would only bring him up to 3MM.  If he hits the 5% chance of being at least a .500 pitcher, he should be getting way more than 3MM.  That Prior settled at 3MM shows how he himself didn’t think it was even worth it to debate it.

Mark Prior got Willie Bloomquist money.

Now, the Padres aren’t going to come out and say “we believe he has a 95% chance of playing no better than what we might be calling up from the minor leagues”.  But, giving him a 1yr/1MM deal is doing exactly the same thing.

Fans will obviously think “Mark Prior”, and figure the odds at 10%, 40%, 40%, 10% respectively, and figure that he could have signed a 1/6 deal, and they’d be happy.

Padres say: R.I.P. to the ballplayer once known as Mark Prior.

Note: I have no idea of what his forecast should be.  Marcel sees him as a .470 pitcher, with 7 full games, or a WAR of 0.6, a salary of 3MM.

I thought the exact same thing, regarding the upper-end of his salary possibilities.  I’m shocked that no other team offered him a better deal.  I mean, with a $1m base and the “right” incentives thereafter, you can’t lose.  How do the Yankees not get involved here and throw $1m base at him with incentives up to $10 or $15m?

Incentives are based on playing time, and since Prior probably won’t be available until July at the earliest, the very best case scenario meant only getting a half seasons worth of performance. It would be extremely difficult for any healthy player to be worth $10 million in that time span, and no way could he possibly deserve $15 million.

I’m guessing that Prior realized that there was no point in trying to fight for a large contract this year and instead decided his best bet was to focus on getting healthy and get a bigger paycheck next offseason. If he can just end 2008 in one piece with decent numbers he can reclaim some of his value, and going to a warm weather city with a large pitchers park in the NL is the best way to achieve that.

But they’re not based only on playing time, right?  You don’t have to have the player calculating his WARP after every start, but you could set some kind of tiering system wherein if he has under a 5 ERA and over 5 wins, he gets $2m.  Under 4 ERA and over 10 wins, $5m.  Under 3 ERA and over 15 wins, $10m.  Or whatever.  It’s arbitrage - figure out how much he’s worth at various performance levels, and offer incentives such that he never earns what he’s worth.  You can’t lose.

I didn’t realize he wasn’t coming back until July, I guess that explains part of why the upper end of the contract isn’t that impressive.  And there might be something to the whole NL pitcher’s park thing, I was just thinking that.

Stuff like this just baffles me though.  We have teams more than willing to pay $10m per year for 4 years for league average pitching.  Pineiro got $4m to be middle relief after posting a 6.00+ ERA the year before.  Is Prior’s outlook really that much dimmer than Pineiro’s?

Prior was born in San Diego, went to high school there, and went to college at USC.  This may play a large part in his choice to sign with the Padres.

If Petco was 5280 feet in the air, no way he goes there!  Very interesting about his background.  Add in the pitcher-friendly park, and it would have been a shock that he not sign there.

Man, what a lottery ticket he hit.

Erstad: paying like Prior.

Marcel has him as a terrible hitter, -2 wins.

The Fans see him as an average-fielding CF.  UZR still sees him as a very good CF (+1 win), as does Dewan.

Positional-adjustment is +0.5 wins.  Aging already included.  Replacement-level for AL stats is 2.5 wins.

So, his WAR per 162 is between 1.0 and 2.0.

Marcel has him forecast for 54% playing time (which jives with a 1 WAR evaluation).  UZR/Dewan would argue for a full-time role, but he’s essentially the same as Michael Bourn.

His final WAR is somewhere between 0.5 and 1.5.  Any way you slice it, this is a good signing (value of 2.6MM to 7.0MM, depending what you think of his fielding).  People will *always* focus on his very poor hitting, and dismiss the value he brings with the glove.

Tango/318 - You say that aging is already included. Does that include aging his fielding projection? And do you have any data that says fielding and hitting follow the same aging curve?

From MGL:
http://tangotiger.net/mgl/agecurve.pdf

Go to page 4.  From age 25 (-2 runs) to age 35 (-11 runs), you lose about 1 run per year.

From age 30 (-4 runs) to age 36 (-17 runs), you lose about 2 runs per year.

So, depending what you want to call it, expect to lose up to 2 runs per year.

(Don’t look at each data point too much, since it was based on just a few years of data.  Try to smooth it out in your mind.)

Oscar Villarreal: 2/2.85, paying for a total of 0.52 WAR over 2 years.  5 years of MLB service.

Marcel sees him as a .520 reliever.  Replacement level is .480 in the NL, and he gets 8 full games.  So, he’s 0.32 WAR in 2008, and 0.22 WAR in 2009.

His 0.32 WAR, at 80% of free agent value, implies 1.51 MM.  He’s young, so I’m not sure what to do with his 2009 season.  Applying the standard aging curve, and calling it 0.22 WAR gets him 1.44 MM at full free agent value.  Total value is 2.95 MM.

He signed for 2.85MM.  Bingo.

FINALLY.  A reliever signing that makes sense.  And he’s young to boot.

JC calls Rowand the “deal of the year”:
http://www.sabernomics.com/sabernomics/index.php/2007/12/the-deal-of-the-year/

I disagree.  Trying to give Rowand some break here, I can call him as +1.5 WAA per 162G, and giving him 84% playing time as Marcel suggests, and being in the NL, gives him a WAR of 2.9.

His 5/60 deal has an implied WAR of 3.25.

I definitely don’t see this as the “deal of the year”.

"But they’re not based only on playing time, right?  You don’t have to have the player calculating his WARP after every start, but you could set some kind of tiering system wherein if he has under a 5 ERA and over 5 wins, he gets $2m.  Under 4 ERA and over 10 wins, $5m.  Under 3 ERA and over 15 wins, $10m.  Or whatever.  It’s arbitrage - figure out how much he’s worth at various performance levels, and offer incentives such that he never earns what he’s worth.  You can’t lose.”

Incentives can’t be based on performance (stats). Only playing time (plate appearances, innings pitched, or games finished for relievers) and I think a few end of season awards. With that in mind, the very most it probably makes sense to offer him in incentives is $6 million.

Curt Schilling’s fat ass begs to differ.  Where did you read that there were only certain things that incentives could be based on?  I think ARod’s new contract is supposed to have some kind of incentives related to him breaking certain milestones like the all-time HR record.  Maybe that’s worded in some bizarre way that conforms to the rules as you stated them… (I know Manny has a no-trade clause that gets activated if any other teammate has a no-trade clause, but somehow Varitek got a no-trade clause without kicking in Manny’s… so I certainly don’t mean to underestimate agents’ ability to do weird stuff with words).  But I thought Schilling’s $2m weight clause - 6 weigh-ins, if reports are accurate - was pretty cut and dried.

Regarding the amount of incentives, I guess if they’re based on playing time only, your $6m cap might make sense.  Presumably he’ll only pitch a full season if he’s better than a replacement-level pitcher (else they’d just cut/bench him and pull up a minor leaguer).  So in order to ensure you’re not overpaying him, you’d have to set the cap on his incentives at the value of a season of replacement-level pitching.  I’m sure Tango has a dollar figure he can put on this.

I love Rowand.  I have him as 1 win in defense.  1/2 win above average in hitting, a few runs in baserunning and 3 runs in position, for a total of 2 wins above average.  That is around 4 WAR in the NL.  While not the deal of the year, I think it is a very good deal.  Which is suprising considering that he had a career year, hitting-wise, in 07.  As we keep saying, a player at a premium defensive position who is very good defensively will tend to get underrated and underpaid.  The reason we can say that with impunity is that players at the corner OF positions and first base who are good hitters (but not necessarily that good for their positions) are the ones who are usually overpaid, especially if they are bad defenders (see Lee, Manny, Konerko, et al.).

A. Everett still has to be the deal of the year, by far.

Everett was a 5+ service years player.  Not sure why they reported he signed a “Free agent contract”.  He wasn’t a free agent.

Anyway, it was only a 1yr deal, and at 80% of free agent value, if he’s a 2.5 WAR player, thats 9MM of value for 2.8MM.  Certainly, a great deal.  They are paying him as a 0.7 WAR player, so certainly a massive difference in value.

Typically, we’ll be off within 0.5 wins between us and MLB management.  When we are off by 1.0 wins, it’s a huge difference.  A 1.8 win difference is gigantic.

None of the other position players have been great deals.  They’ve all been pretty much par for the course.  Maybe Milton Bradley, but he carries alot of uncertainty.  I’m still waiting for Pedro Feliz.  He’ll probably sign the worst deal for a player / best for management.

***

MGL, you have Rowand as +2 WAA.  Is that after aging or before?

Everett was a 5+ service years player.  Not sure why they reported he signed a “Free agent contract”.  He wasn’t a free agent.

Everett was non-tendered, thus making him a free agent.  At least that is how I thought the process worked.

Ah, I see.  So, he was non-tendered, and any other team could have bid for him, and didn’t? 

That’s ALOT of teams asleep at the switch.  That pretty much tells me that sabermetrics has outlived its usefulness to MLB.  We’ve done as much as we can.

R.I.P. Sabermetrics.

"That pretty much tells me that sabermetrics has outlived its usefulness to MLB.  We’ve done as much as we can.

R.I.P. Sabermetrics.”

Somewhere, Gary Huckabay feels vindicated.

I have a question that Everett going to the AL made me wonder about, and it might not make any sense but this is the place to ask.

First, I realize that right now the AL is better than the NL. I understand that adjustment, but that’s not really what I mean here. So I guess for the purposes of my question, I’m assuming that both leagues are equally talented.

Would a player who is so unbelievably awful with the bat be better served in the AL (where he could bat ninth, and in theory be the only very weak part of the lineup) or the NL (where he would bat 8th, in front of an even worse pitcher)? Is one dead spot in the lineup in the AL worse than two back-to-back in the NL?

#330, I doubt that it makes much, if any, of a difference.

My projections are always for 08, which “includes aging” I guess.

It doesn’t really matter where you are weak, really.  It’s your overall value that counts.  Sure, there are leverageable aspects to consider, but not much.

Re: Tango #328 - We don’t know for sure that nobody else bid on him, I don’t think. It seems obvious that teams are underrating his defense, but it’s a big stretch to say that nobody else even made an offer.

If Everett were smart, he would have waited more than 3 hours to sign a deal.  I’m guessing he is smart.  It’s shocking really.

Dan/329: It’s not that baseball analysis is dead.  It’s the appliers of baseball analysis that are dead.  In any case, they live in a different world from sabermetrics.  Through the eyes of the movers and shakers, “RIP Sabermetrics”.  Through the eyes of thinking fans, “RIP Movers and Shakers” (at least half of them anyway).

***

Reed Johnson: 1/3.3MM, paying for 1.1 WAR.

He’s a 4+ years of service, meaning he gets 60% of the free agent value.  1.1 WAR means he gets the 400K flat salary, plus 4.8MM for his marginal value.  Take 60% of that marginal value, give him the 400K flat, and you get 3.3MM.

He’s a CF playing LF.  Fans see him as an excellent fielder.  UZR *loves* him in LF (hates him in RF, and isn’t too impressed with him in CF).  Dewan digs him in LF.

He’s at least +1 wins in LF.  You can even stretch that to +1.5 wins.

LF knocks out 0.5 wins on him.

Marcel sees him as a slightly below average hitter, with 400 PA.

Add it up: he’s an average to +0.5 wins overall.  He’s in the AL, so his WAR is 2.5 to 3.0, per 162G.  He gets knocked down based on his expected playing time (60% of games).  That puts his WAR at 1.5 to 1.8.

Reed Johnson had a terrible year with the bat in 2007, plus he was injured.  That’s a deadly combination.  He should have signed a 4-5MM deal.  The uncertainty cost him.  But, it’s only for 1 year.

Steve Swisher:
+1.0 win as a hitter
+0.5 wins as a fielder+position (fans see him a smidge above, UZR loves him in the OF, hates him in the IF, and Dewan sees him a smidge above overall)

He gets 85% playing time.  His WAR is 3.4.  He’s young, obviously.  He’s got 3+ years of service, meaning three more years before free agency (would have been for the 2011 season).  He’s actually signed through 2011 (with club option in 2012).

Let’s give him a 0.25 aging, since he’s still at his peak, and applying 40%, 60%, 80% to his free agent value for the first 3 years.

Through 2012, Swisher Properties will generate 79MM of dividends.  Including the club option, he will be paid 35MM.  (He should be paid 60MM… what a deal the A’s got here.) He’s got a surplus value of 34MM. 

I don’t know about the other players.  However, a single rookie, expected to be a 2 WAR player for each of the next 5 years, will generate 56MM in dividends, but will cost you 17MM in salary.  That’s 39MM of surplus value.

***

Put another way, Swisher is expected to generate 14.5 WAR over the next 5 years, while the rookie of MLB average talent will generate 10 WAR.  Those extra 4.5 wins will cost you 17MM.

***

What I am shocked over is that the A’s would take a player with so much surplus value and spin him for other players of presumably even higher surplus value.  Typically, you would trade guys with limited (or no) surplus value for guys with some surplus value.

Who are these guys the A’s got?

34MM = 44MM

According to Kevin Goldstein of Baseball Prospectus, De Los Santos and Gonzalez were the top two prospects in the White Sox system.
http://www.baseballprospectus.com/article.php?articleid=6912

De Los Santos pitched last year in A-ball at age 21.  According to Goldstein, “His fastball sits at 91-95 mph, touching 98 at times; it also has late movement, and he commands it very well. His breaking pitch is a power curve with hard late bite, and it’s a true out pitch when he’s on. He understands the importance of developing an offspeed pitch, and he improved his changeup over the course of the year.” He’s projected to be ready for the majors in 2009 and is viewed as having high upside potential.

Gonzalez was the White Sox first round pick in the 2004 draft.  He’s a lefty that pitched last year in AA at age 21.  According to Goldstein, he has an average-velocity fastball and one of the best curveballs in the minor leagues.  He is viewed as being ready for the majors sometime during 2008.

De Los Santos had 18 wild pitches in 98 innings!  Seeing that only 80 runners reached base, that’s a heckavu rate!

Since 1993, the worst career rate is Scott Williamson at 13%.  K-Rod is next at 9%.  De Los Santos is over 20%!

The fact that De Los Santos had 18 wild pitches with Kannapolis (with ~89 baserunners) and only 1 wild pitch with Winston-Salem (with ~25 baserunners) leads me to believe the catcher and/or scorer had some effect on the WP stat either at Kannapolis or at Winston-Salem.

If you look at the other members of the Kannapolis starting rotation, Rasner had 12 WP, Edwards 10, Zazueta 11, Long 4, and Perez 6.  Several of those are high, but none as high as De Los Santos.  The primary catcher for Kannapolis was Francisco Hernandez.  I can’t find any scouting reports on his defense other than that he has a strong arm.

I would put Swisher closer to 2 wins as a hitter.  He’s +23 and +27 the last 2 years by B-Ref, about average his rookie year but he’s still young.

I don’t know how to value the guys the A’s got.  I don’t project either to be league average in 2008, and with young pitchers the odds are only one of the two will have a significant major league career.  Maybe they have the same value as two 28 year old average pitchers, maybe not.  They are pretty much lottery tickets at this point.

"Break” discussion has been moved to here:
http://www.insidethebook.com/ee/index.php/site/comments/pitch_analysis_of_eric_bedard/

He was born Nov, 1980, which puts him pretty much at his peak right now.  Your numbers (including rookie year and regression) would suggest +1.5 wins.  Marcel is unaware of parks.

I posted this on a Whitesox board:

Swisher had great UZR in the corners and terrible in CF (25 run difference).  This is fairly inconsistent with the typical player who plays the corners and center (10 run difference).  Furthermore, Swisher has average tools across the board, nothing that would explain such a gap.

Quentin was average in the corner.

Tools-wise, Quentin is slightly better than Swisher.  If you have to have both fielders out there, I’d put Quentin in CF and Swisher in one of the corners.  If I could, I’d flip-flop Swisher/Dye based on who’s on the mound, or preferably, who is at bat!  It would be the ultimate platoon.

tango/#336

It’s 2008.... Steve Swisher retired a ways back

Now back to your regularly scheduled programming.

Ouch! 

I didn’t realize (or remember) that Steve is Nick’s father.

#344 - On switching Swisher and Dye I wonder if your gains in having the better fielder in the position more likely to get a ball hit to him would be outweighed by not letting them get comfortable in one spot.

As for switching them between batters, I’m pretty sure you’d just wind up with 2 pizzed off outfielders.  Then you’d get MLB to crack down because if delays the game a bit.

Whitesox did it for one inning in the 9th inning of a close game.  I think it was Sammy Sosa and Lance Johnson switching back and forth.  They were having a ball running back and forth.

I’ve seen some extra inning games where a pitcher winds up in the outfield and switches back and forth with the real corner OF in a game.

I can see it being fun as a novelty, but getting real old if it were used everyday, especially for an old guy with bad wheels like Dye.

I agree it would be tough on morale.  As Peter Gammons likes to point out: players are humans.

But, when I was playing company softball, we’d do this ALL the time.  We moved the 2B/SS back based on the batter.  We were professionals about it.  This happens in the corporate world too.  We are humans.

It’s ballplayers that have egos that are as fragile as glass.  Ask Scottie Pippen.

Zumsteg’s post made me retake a look at the “10%” figure I’ve been using.  Maybe 8% would be better.  Here’s what I wrote:
http://ussmariner.com/2008/01/06/silvas-contract-free-agency-my-general-economic-pessimism/#comment-247009

Team payroll, franchise valuations, and MLB revenues, have increased at an annualized 10% rate, if you look at the last 20 and last 10 years, and probably last 5 years.

I expect all three of those to continue the 10% rate. In any case, MLB teams expect that, since what they are paying out for free agents has those implied assumptions. (Otherwise, my free agent tracker would be wrong.)
...
Ok, I’m looking now, for the 1990-2007 time periods.

1990-2007: implies a 10%
1996-2007: implies a 9%
2003-2007: implied a 4%
2006-2007: 6.5%

So, it depends where you grab your boundaries, obviously.

So, 8% might be a more reasonable figure.

Franchise valuations are alot more stable. Using the same time periods as above:
1990-2007: implies a 6%
1996-2007: implies a 9%
2003-2007: implied a 8%
2006-2007: 12%

Again, 8% might be more reasonable.

It’s important to note that team valuations are somewhat inversely proportional to player salaries. After all, if you can reduce your expenses, while keeping the same quality and revenue, your profit margin will go up.

They seem to hover around 8%.

Working against the players is that there’s more revenue sharing (meaning less likely for a team to see increased revenue for increased performance… since that revenue will be siphoned off).

Working against the players is MLBAM which is pure revenue sharing.

Working for players is that owners treat the extra capital appreciation as a bonus, and might give it to players.

I’ll have to look at the implication of the 8%, and if this throws off all my numbers.  It’s possible that I need to set the growth to 8%, but bump up the starting $ per win a bit to compensate.

Brewers signed Cameron for 1 yr 6.22M assuming they buyout the 2nd year or 2 yr, 15.47M if they do not buy it out.  Keeping in mind he misses 25 games to start the season.

Not sure you add this but it also moves Braun off of 3B and Hall out of CF which has to be worth 2-3 extra wins defensively.

In post 66, I have Cameron very high.  I love that guy.  Everyone should.  Way above average.

Signing him to a 2/15 deal is a steal for the Brewers.  Sending Hall to 3B, and moving Braun to LF is exactly what I called for.

Brilliant.

Tango/353: Checking the data on Sean Smith’s website (ZR and TotalZone), Cameron is between a -2 and a -7 for 2007. I don’t have access to the UZRs like you do, but the Hardball Times ZR data is the more pessimistic about Cameron, and that’s the same data set Dewan uses, I though. Am I missing something here, or did Cameron just have an off year defensively? Does that change your opinion of the signing if he did have a down year defensively?

Over the last 2 seasons, Cameron’s UZR was +15 runs per 162G.  Fans still see him as above average.  Dewan still has him as above average.

As a general rule, quoting one season of data as indication of talent is grounds for banishment from this site!

Which raises the question: how many posts do we need to judge someone’s true posting talent level?

I call “scrappy, gritty” white guy flirting with replacement-level.

Yeah, really spent enough time in my life wandering in the desert already.

If we’re trying to predict his future defensive skill, though, shouldn’t we weight his ‘07 performance more heavily than his ‘06 performance, similar to the way the Marcels work on offense? Again, looking at the data on Sean Smith’s website, his outfielder projections are +3 as a CF.

Is this just an issue with the specific dataset here?

You definitely would want to weight more recent seasons more, if for no other reason is that your body and reflexes are more similar to what they were a year ago than 3 years ago.  And the further away you are from age 27, the likely more weight you want to place on the most recent season.

As for the differences implied by the different measures, it all depends on the level of granularity of that data, and the various adjustments made to establish the context.

And even if you have two guys forecast for +3 as a CF, one Mike Cameron and another, I dunno, Steve Finley, then you must revise your forecast for the two players, since you have additional knowledge of the data.

I think part of the issue here is regression to the mean - to oversimplify here, for the sake of illustration, there are two kinds of outfielders: guys with speed and guys without speed.

Certainly we should expect Cameron to perform better than the Finley types because we know he’s not a statue. And his arm still seems to perform well (I checked a revised version of my arm ratings from baseball-fever.com, and his ‘07 performance seems to be one of the best in center field).

Realistically the best thing to do would be to have a PECOTA-esque defensive projection using some sort of similiarity scores. But that seems to be a long way off.

Not way off at all:
http://www.tangotiger.net/scouting/sim2007_5492.html

Among those OF, I’d guess their average UZR would probably be on the plus side, so that’s a better regression point than zero.

I did one for Jose Guillen a few weeks ago.

Cameron is a steal.  I have him as .5 wins in UZR (in CF), and even in hitting/baserunning and arm.  That makes him almost 2 WAR.  A steal, even at his age.

Yeah, you regress a player’s UZR to a similar player in age and speed.  My recent article on speed and UZR suggests that an average fast player (for his position) in the OF is something like +3 or +4 in UZR and a slow guy is like -5 or even more.  So it is very important to regress a player’s sample UZR’s using his speed rating.

Tango, a few things.  I was surprised that Edmonds was rated as a better CF defensively by the fans than Cameron...if he can still put up a good OBP and have a SLG in the .400’s, it doesn’t appear like the Padres will lose much going from Cameron to Edmonds.

I was surprised that the Padres didn’t offer a similar type contract to Cameron (I live in San Diego now, and the newspapers were saying that Cameron’s price was too high).  The Padres have a good front office in place and $4.22 million for one-year plus a team option (which they wouldn’t use) would seem to be a good deal for the team.  Maybe they were concerned about him missing 15% of the season…

Fans may be biased with Edmonds.  When it comes to players in their late 30s, I start to discount Fans’ ratings heavily.

So my stats have him at +3 and MGLs at +5, not much difference, though the stats I have are much more crude.

Steve Finley was at one point an outstanding defender.  By his late 30’s he was useless in the field, and won gold gloves anyway, and then at 40 he went to the Angels and everyone who didn’t believe UZR in 2002-2004 had a wake up call.

Cameron’s 35.  I doubt he’ll be much of a defender if he plays as long as Finley, but as a one year deal, I think the Brewers did very well.

MGL’s article about regressing to speed scores was a great one.  I did it for my projections, so that would partly explain how my projections are pretty close to his despite not using any data I can’t get for free.  Its a great thing to have in considering defensive projections for minor league outfielders who might get a chance to play.

Tango/360: Yes, but we don’t know how well any of those players performed in 2008 because, well, it hasn’t happened yet. Ideally we’d use sim scores to find similar fielders, and then use their performance from the next season to build our aging curve, a la PECOTA.

Yeah, it does seem that a lot of OF’s who are deemed “great defensively” get that personal bias sometimes when rated subjectively.

Rally, I don’t know if MGL regresses to the speed scores in his UZR even with the article...I know he listed sample size issues, etc.  Either way, it’s a step in the right direction towards not just regressing to zero (I know we talk about regressing towards height, etc. but I still don’t think any forecaster does that).

From post #361 it looks like MGL regresses UZR to speed scores.  I regress my outfielder defensive projections to speed score.  In my CHONE projections I regress power to weight (from what I’ve seen height doesn’t make much difference) and BABIP to speed scores.

That brings up another good point...we need to know what statistics and such regress to what subjective characteristics if we don’t want to regress to zero.  Do we regress defense to speed score, height, weight, age, round drafted, etc. or just speed score?  That goes for linear weights, UZR, etc.  There are a lot of different things we can regress to (I only mentioned a few above) and we may not need to use them all for each compenent we’re regressing.  In other words, different components will have different variables that we regress to (other than zero).  I don’t think anyone has 100% determined what those are for each component, but it seems like we’re moving in the right direction.

I use height and weight to regress different components of my projections.  Pecota claims to do so as well, although their whole methodology is a black box.

Determining what to use for regressions is not all that difficult, I don’t think. The first step is deciding what might be relevant.  The second is to run correlations and what have you to determine relevance.  For example, I doubt that anything much correlates with defense, at least for the outfield, other than speed.  If height and/or weight do, it is most likely related to speed, in which case we ignore height and/or weight and just use speed.

That’s about all I do.

In projecting power (and other components), it is important to use physical characteristics.  If Prince Fielder or Ryan Howard hits 40 HR in his first season, that is A LOT different, in terms of a projection, than if Brian Roberts or Jimmy Rollins does the same thing.

What’s the consensus on the Rolen/Glaus deal?  Both guys are very hard to project due to recent injury woes. If both are relatively healthy, Rolen may be the better gamble because of his defensive value.

I think there’s a place in baseball for medical sabermetrics, combining what you guys do, with an in-depth case study and knowledge base, to properly project the effects of different types of injury. Any bored Physicians or Re-hab Physiotherapists out there with the requisite analytical abilities to do this kind of work?

Nashboy, I don’t think there is anyone out there that can do what you’re asking.  In the Rolen’s/Glaus deal, Rolen’s obviously is one of the best 3B defensively (and Glaus is not very good), but it seems unlikely that he will play as much/more than Glaus over the next two years.  His injury, according to reports, seems to be far worse than Glaus’s.  Also, Rolen’s is slightly older and is being paid more money in average annual salary.  I don’t think anyone can accurately project what is going to happen to each player over the next two years, but a doctor could tell you the likelihood of the injury recurring, etc.

FWIW, Glaus will take a hit moving from Toronto to St Louis.  Toronto is one of the American League’s friendliest power parks for right handed hitter, while St Louis is the opposite.  Just something to keep in mind.
vr, Xei

Some of that will be balanced by Glaus facing weaker competition in the NL.  Rolen has more guaranteed money left on his deal, but if he’s healthy that’s to Toronto’s advantage, a healthy Rolen is worth more than the 12 million per year he’s paid.

Whoever wins this trade is going to come down to a question of health.

I have them roughly equal in a projection, not account for league differences.  Given the league differences, the fact that Rolen is slightly older, makes more money (I think), and has had the worse injury, I think that STL is the clear winner.

Of course, a player like Rolen would get leveraged behind a ground ball staff with two (or more) lefty starters…

Rolen v Glaus

Marcel gives Glaus an advantage on offense (.360 wOBA v .336), with both getting close to 500 PA.  That’s a 1 win difference over 500 PA.

UZR has a 1.5-1.8 win advantage for Rolen, whether you look at 2006-07, or 2003-07, for 115 games.

There’s a 0.5 win difference in league baseline.

Overall, a somewhat equal exchange in talent.  Basically, there are both around a 2.25 WAR.

Rolen has 3 years, 36MM left on his deal, plus 10MM deferred.  That’s 46MM outstanding.  He should be getting paid 26MM.

Glaus has 2 years, 24MM left on his deal.  For his 2 years, he should get 19MM.

The Cardinals are giving the BlueJays 4MM, which basically means that the Jays are paying 42MM for 26MM of value, and that the Cards are paying 28MM for 19MM of value.

If you want to bump up Rolen to a 2.5 WAR and Glaus down to a 2.0 WAR, that makes Rolen Properties worth 30MM and Glaus Properties worth 17MM.

That puts the Jays at overpaying for Rolen at 11MM and Cards at overpaying for Glaus at 11MM.

Presuming the injury risk cancels out, that’s about as close a one-for-one, same-position, same-injury-risk trade that you’ll find in baseball.

"# $10M deferred ($2M/year from 2003 to 2007)
# full no-trade clause (waived to allow trade from St. Louis 1/14/08, with Toronto paying St. Louis $1.8M and the Cardinals responsible for $4M bonus due in 2010 as part of the deal)”

The above per Cotts. I’d say no way does Toronto assume the $ 10M deferred rung up by the Cards. The $1.8M appears to be 3/8 0f a $5M signing Bonus paid to Rolen back in 2003 ( the Jays do this kind of thing with bonus money because of arcane Canadian tax reasons). St Louis is paying the back end $4M bonus. So really it’s 36-37.8 the Jays are on the hook for.

How do you guys account for injury in your projections? Case studies of recovery times of players with similar injuries? Are all injuries treated the same?

I did 12x3 + (10-6) = 44 as to what the Jays are on the hook for.

I don’t know if I’m right or not.

As for Marcel, it’s extremely basic in forecasting.  You can read about it here:
http://www.tangotiger.net/marcel/

re: injuries

Pretty much the same way Marcel accounts for injuries.  Projected playing time is a function of recent playing time.  Rate stats are affected to the extent that htey affected the player in recent years.  Nothing fancy.

The Texas Rangers signed Jason Jennings to a one-year deal worth $4 million. Before I did any math, I figured this would be a steal as teams would overvalue 2007 and undervalue 2006 in their evaluations. Jennings is projected as a .485 pitcher in 130 IP or 14.4 complete games (rounded avg. of Chone and Marcel). I have him as a 1.4 WAR pitcher ((.485-.390)*14.4) which should pay about $6.15 million. So, as I initially suspected, this comes out as a terrific deal for Texas.

Jennings is also an excellent hitting pitcher so he’d be worth a smidge more to an NL team.

What are the thoughts on the Lieber (1/3.5 I believe) deal and the Robertson extension (3/21.25)? Kind of surprised at the Robertson deal since they’re locking up a pitcher who is mediocre at a price that doesn’t seem like much of a discount. As for Lieber, he’s not guaranteed a rotation spot but at that price (and for a team that can certainly afford the financial risk), he seems like a fine guy to take a chance on at a reasonable cost.

What would you guys say about the proposed Santana to the Mets deal? I remember you said that Johan would be worth $28 million in today’s market, and he’s getting paid $13MM. The Mets would then have to sign him to a massive contract, but there is a good likelihood of that contract being under-value, as no pitcher has ever been paid nearly as much as Johan is worth.

How would you go about obtaining a likely dollar value (in free agent dollars) for the prospects the Mets would surrender, for that same time period? It would be especially useful to weight those dollar values according the boom/bust possibilities. (Maybe that’s included in the projections on unknown commodities?)

On the prospects,

If I were a subscriber, I would look at PECOTA’s percentile values.  Determine how much the player is worth at his 90%, 80%, ...50%, etc.  Then take a weighted average of those values.  With established vets you can just look at their 50% mean, in other words what CHONE, ZIPs, Marcel, etc. will give you.  ZIPs is ahead of the others in that it gives you percentile values for some players (and Dan could do it for other players if he wanted to).

I think you have to do it that way for young players because their 50% mean is quite likely below a replacement level, especially if he’s not yet a AAA player.  Even if you project 3-4 years of improvement to them, the mean of a non-elite prospect still might not be much better than replacement level.  But take 3-4 prospects like that, odds are one of them will develop into a better player and have some real value.

John/381: asked and answered here:

http://www.insidethebook.com/ee/index.php/site/comments/how_much_is_troy_tulowitzki_worth_in_a_trade/

Please continue commentary on that thread.

Rally/382: yes, the percentile forecasts is the ONLY place where they make sense here, since there is a floor to the value of a player (zero dollars).  That said, I doubt you even need to go there.

I guess you could come to the same estimate by answering one question:  What is the % chance this prospect is a league average player?

Carlos Pena: 3/24, paying for 2.5 WAR.

Pena has 4+ years of service, meaning he gets 60% of his free agent value in 2008, 80% in 2009, and a free agent in 2010.  If he were a free agent, he should have expected 30MM, if he agreed to 24MM under his current service time.

Hitting: Marcel loves him as a hitter, +3.0 wins per 162G.

Fielding: Fans like him (though there may be some question as to the Tampa ballots), Dewan sees him around average for 1B, and UZR sees him as below average.  Hard to call it here.  Let’s settle for average.

Aging included in hitting, not in fielding.

Position is -1.0 wins.

Repl level boost, for AL, is +2.5 wins per 162G.

Marcel is forecasting 73% playing time.

Adding it all up:
0.73 * (+3.0, +0.0, -0.2, -1.0, +2.5) = +3 WAR

Looks to me like Tampa got a good deal here.  A fair deal would have been 3/30.

Tango/386:

You say that a “fair deal” would have been 3/30, and this implies that Pena somehow got ripped off. But look at it from his perspective. Tampa is his 5th major league team and he doesn’t even have 5 full years of service time. He’s still relatively young, but he’s close to being a journeyman. A 3 year contract is security that he was probably welcoming (and also probably not expecting a year ago). So while he might have not gotten market value in terms of dollars, isn’t 3 years of guaranteed money more than fair for a player with his track record?

I’m already discounting his 2008 worth by 40% and his 2009 worth by 20%.  So, his 3 year guaranteed contract is in comparison to other expected 3 years guaranteed contracts with the 3rd year being a free agent buyout year.

Now, we don’t have exactly the same situation to compare to, which is why we create a reasonable model.  And that’s what we have here.

However, it does seem that long-term arb-eligible deals have been getting a further discount than my year-to-year discounts would expect (see: Utley, Mauer, et al).  That is, while year-to-year, they may deserve a 40% and 20% discounts for their last 2 years of arb, it seems that signing a 3yr or longer deal has a further 20-35% discount on top of that.  And it seems that this discount (for long-term deals) exists for the arb-eligible player, because those guys are really valuing security.

Big names avoiding arbitration, with service years in parens:

Holliday: 2/23 (4), paying for 4.3 WAR
Cabrera: 1/11.3 (4+), paying for 4.1 WAR
Morneau: 1/7.4 (3+), paying for 4.0 WAR
Teixeira: 1/12.5 (5), paying for 3.4 WAR

Kazmir: 1/3.8 (3+), paying for 1.9 WAR
Lidge: 1/6.35 (5+), paying for 1.7 WAR
Street: 1/3.3 (3), paying for 1.7 WAR
Crede: 1/5.1 (5+), paying for 1.3 WAR
Capuano: 1/3.75 (4+), paying for 1.3 WAR
Downs: 3/10 (5+), paying for 1.2 WAR
Ramirez, Horacible: 1/2.75 (5+), paying for 0.7 WAR

If someone wants to figure out some of these guys, feel free.

What gets me is Horracible Ramirez. Marcel has him for 13 full games.  At 0.7 WAR, that makes him about +.055 above replacement.  In the AL, replacement is .370 for starting pitchers, so that puts him around as a .425 win% pitcher.

Why in the world would you dip into 80% of free agent value to get a pitcher like this?  Surely the Mariners must have plenty of pitchers in the minors under their control that can be a .425 pitcher?

I mean, the reason you dip into the free agent pool, and pay double the worth rate is that, is because you need to shore up a hole.  But, if what’s really available to you is a bunch of .425 pitchers in the minor leagues, it’s insane to pay Ramirez anything more than 400K.

I don’t get it.

In regards to Ramirez, I think it’s a combination of two things:

1)the org is really embarrassed about giving up Rafael Soriano for him and so they are hoping that he does something, anything to justify the trade

2) they see something in him that makes them think he can be decent. What that is I have no idea.

And yes, the M’s have lots of guys who are better than him who cost nothing. If they don’t trade for Bedard they have at least a half dozen guys who could fill the 5th spot adequately. If they do trade for Bedard, then HoRam is simply a very expensive long reliever, and there are even more guys who could do his job just fine for a fraction of the cost.

Does anyone else enjoy the fact that Ed Wade are paying Ty Wigginton 4.4M compared to Adam Everett’s 2.8M? 

Makes me wonder if either a) they don’t understand the value of Everett’s defense or b) Tejada didn’t want to move to 3rd.

Tango/390:

Not that this necessarily makes logical sense (when taken to the extreme of giving free agent money to a terrible pitcher), but I think the thing you aren’t getting is that Ramirez is left handed. I really do think there’s some truth to the idea that as long as you throw with your left hand, you can get a job somewhere in pro baseball.

re carlos pena: Marcel is projecting a 73% playing time based on solely his major league playing time, correct?

This will be a horrible under-estimate because pena has been missing games in the majors not because of any lack of durability but beacuse teams wouldn’t give him a job.

I think the deal for Tampa is markedly better than you suggest.

The exact formula for all players is:

PA = 50% of yearX + 10% of yearX-1 + 200

For Pena, that’s 612 PA last year and only 37 in 2006.  That’s 510 PA.  If you give him his next career high (561 PA), that gives us a PA estimate of 562 PA, or 80%.

Note that even guys who have 700 PA year after year are forecasted for only 620 PA (90% of playing time).

Under that scenario, he’s close to a 3.5 WAR, making this one of the best deals of the year (for a team).

Pena actually had 543 pa in 2006 split between majors and minors, which gives you a practically identical projection of ~560 pa. Zips is in agreement as well, forecasting ~570 pa. 80% playing time and 3.5 WAR looks good, a nice deal.

Question: Do you use that formula for Pitchers as well as position players? (appreciating that playing time projections are somewhat problematic to do based solely on previous playing time)

I don’t remember exactly what it is, but if you go to the Marcel page, and follow one of the links there, I describe everything in detail in one of my blog entries.

But I seem to remember I do the same scheme.  You can double-check with a couple of guys.  I think I set the baseline to 60IP for a starter and 25 for a reliever.

Using a Marcel playing time projection for lots of players makes little sense.  As will points out, you need to know why a certain player did or did not play in any given year.

Sometimes you have a model that works in the aggregate but is useless when you have just a little bit of information.  In the case of projecting playing time, you usually have information that trumps the Marcel model.

I would be willing to bet that a person can beat the Marcel playing time projections at least 75% of the time.

For example, if a player has not played much because he was a part-time player for whatever reason and the team announces that he is now full time, what is the value of the Marcel?  If a player missed an entire year because of a catastrophic injury like a broken bone and is now healthy, again, what is the point of the Marcel (granted, they are probably a risk for another injury or poor performance translating to less playing time).  Etc.

Bascially, I would put little stock in the Marcel playing time estimates, with all due respect to Tango of course.  I’m sure they work just fine with players in “normal” situations (463, 512, 387 PA, for example) or for catchers and things like that.

Pena is a below average defensive first baseman who can hit real well.  I have him as 29 runs abvoe average in hitting per 150 games, 4 runs below average in defense and baserunning, and of course -1 win for position.  That is around 3.2 wins above replacement for me for 150 games (630 PA).

"Pena actually had 543 pa in 2006 split between majors and minors”

Being an Angels fan, I own three rally monkeys.  Since Marcel was the original rally monkey, I asked them to put me in touch with him.  I asked Marcel about Pena’s minor league time, and he responded “what are minor leagues?”

No matter how much I tried to explain, he didn’t seem to get it.  Yet still, his projection is only 2 runs off of MGL’s state of the art projections?

How does the little guy do it?  Part of the monkey’s good fortune is that almost every time you say “Marcel, you should consider X, which moves a player’s projection in one direction” the monkey, were he more industrious, could respond “but if you consider X then you have to consider Y, which moves it in the opposite direction”

In this case X is his minor league playing time, but Y is Pena’s 2006 minor league performance, where he only hit 278/383/490 - my MLE has that as worth 251/351/431.

The monkey comes out OK by ignoring both.

Tulo: 6/30 (1+ years of service) paying for
1.5 WAR, no aging
OR
2.0 WAR, 0.12 wins lost per year for aging
OR
2.5 WAR, 0.25 wins lost per year for aging

Wow.  What an incredibly horrible deal for Tulowitzki.  A BELOW-AVERAGE player, who would stay below average for 6 straight years, would have gotten a 6/30 deal, if he went year-to-year.

There is a risk of course of attrition, at which point he gets nothing at some point.  Or, a risk that he has a great year, at which point he doesn’t cash in.

But, Tulowitzki is a monster.  To discount his performance as much as he did is about as ridiculous a contract (for a player) that you will find.

Anyone like Affeldt as a $3M starter?

#401, nope.

Rally, #399, I don’t really get your point.  Of course Marcel will do as well as a system that takes into consideration X and Y, if X and Y cancel each other out!  No duh!  That is like saying that a system that projecs everyone to be a league average player works great as long as the player is a league average player.  (Or a broken clock keeps perfect time twice a day.) But what if they don’t?  What if Pena had had back to back great, or even good, minor league seasons in 06 and 07?

And I don’t use minor league stats below AA in my “state of the art” system.

What does a basic Marcel project Pena in lwts per 150 games?

mgl/403,

I took Rally’s (delightful) comment as an explanation for how Marcel is able to perform so well against more sophisticated systems, year after year.

In this case, he points out that two things effectively canceled each other out when trying to come up with Pena’s projection.  And that this kind of thing happens often enough that Marcel is pretty close most of the time.

Clearly, there will be cases in which he’ll miss because he doesn’t consider all available information, or take steps to more carefully interpret the information he has.  Your (and others) system will outperform it at those times.

Obviously, you know all of this--but I don’t think the comment was directed at you.
-j

Pretty much what J said.

I certainly wasn’t trying to criticize you, or anybody else.  Just a case where the monkey comes to the correct conclusion while ignoring a lot of relevant data.  And he seems to do it a lot more often than the broken clock.

I didn’t take it as a criticism at all.  Again, to say that a system that does not take important information into account works most of the time because most of the time that information “cancels itself out” (which is basically true) is both tautological and just a pointless (with all due respect) comment to make.

If you were simply pointing out that that is why Marcel works almost as well (or as well most of the time - however you want to couch it) as the more sophisticated systems, that is fine.  That applies to almost anything and everything that is simple and basic (that it works much or most of the time), no?  OPS, ZR, etc.

I think if you think about it, it’s obvious.  ZR works pretty well compared to UZR and OPS works pretty well compared to LWTS.

It just seems that people seem to find it impossible to believe that Marcel can do anywhere as well as it does, compared to PECOTA, ZiPS, and Chone among others.  Marcel is easy to understand, and it is NOT a black box.  Incredibly, those two things are held against it!  When you read reviews or discussion of forecasting systems, Marcel is typically an afterthought, and I think for those two reasons.

(I don’t take it personally, since I’m only a trustee.  I just put a number down as a baseline against which all else should be judged.)

And the fact of the matter is that Marcel holds its own very very well.  This is confirmed by Nate Silver’s analysis and Rally’s analysis.  Those two guys, owners of PECOTA and Chone, know how very little real advantage their systems have over Marcel.

But the common fan doesn’t think about it much, and therefore dismisses it.  Rally is pointing out that all the little things that their systems use “cancel out” alot, which really means that their extra things really must be very little as it looks like things just pretty much cancel out.

Hey, I’m the first one to say that my “complex” system is maybe an eyelash better than Marcel.  Then again, as Tango himself likes to say, there is no reason NOT to do some things that you KNOW are helpful, even if they don’t add much to the overall value.

Plus, I always say (or at least think) that it is foolish, and can get you in trouble, if you DON’T do the proper adjustments for the 5% (or 1% or whatever is is) of the time when you really have to. For those times, using your simple system that works just fine 95% of the time, will make you look foolish.  For example, let’s say that a player has played in Colorado for his entire career and now you want to project him neutrally (in order to compare him to another player) or he goes to the Mets and you want to project his stats with his new team, using a Marcel with no park adjustments is going to make you look foolish.  Of course the proper thing to do is to say, “Don’t forget that my projection system does not use park adjustments, so you will either have to take this projection with a large grain of salt or redo it using some kind of park adjustment.

Right exactly.  I come out and say that Marcel has specific limitations.  And the reason I don’t even bother doing any tweaks is that there is not enough player movement to make it worthwhile to do additional adjustments at the league, park, or even minor league levels.

If there was enough movement