THE BOOK--Playing The Percentages In Baseball

Dewan has Rollins at +7 for 2007.  Rally’s combined has him at -2.  They both agree that Albert Pujols was an absolutely fabulous defender again in 2007 (Rally +15; Dewan +37), and of course Pujols was a much, much better hitter. 

You’ve got to give Rollins a large positional adjustment to have him as better, or else you have to disbelieve the objective indicators with regard to defence.

They are not “objective”.  They are “numeric”, which is a big difference!

They do not necessarily account for all the contexts and biases present.

The Fans are definitely subjective.

As a non-subscriber to WPA, I only see Rollins as being around +1.5 wins on the offensive side.  I think it’s from that viewpoint that people are objecting to the selection.  With your +.5 positional adjustment, he’s up to +2.0 wins.  Chipper Jones (just picking the OPS+ leader), meanwhile, is at +5.0 wins offensively and doesn’t get a positional penalty.  Is the difference between Rollins and Chipper with the glove worth 30 runs? 

Pujols was around +5.0 with the bat, too, loses 1.0 for being a 1B, but likely gets a lot of that back with the quality of his defense.

At any rate, you’re certainly correct that in one week, one will really care ...

His WPA/LI is also close to +3.2.  (Rollins had no clutch score.)

It depends how much the park factor is.

pujols isn’t just the best 1B, but he best or at least one of the best fielders in game. +1 doesn’t seem like enough here.

I actually said +1.5 wins, which is the highest I’d give a 1B.

How much do you think a 1B can possibly add with his glove?  UZR, 03-mid07, has Minky at +13/150G, as the best.  Erstad is second at +12, and Pujols is third at +10. At the bottom end is Fielder at -14, Giambi -11, Jacobs -10, Ortiz -9, Sexson -9.  The Fans data would suggest a range of +15 to -15, which is consistent with UZR here.

I just don’t see how you can reasonably argue much more than this.

I had Rollins at -5 this year in UZR.  He is a little above average from 04 to 06 (2-3 a year maybe), so I don’t think that the -5 is much of a fluke if any.  I seriously doubt that he was a true +20 or even close this year.  He gives the impression of being a great fielder, but I doubt he is.

And, I may sound like a broken record, but using any context-neutral stats (WAP/LI, lwts, VORP, WARP) is a RIDICULOUS way to evaluate someone for an MVP award.  WPA is a little better (maybe a lot better), but the weakness with that, as with lwts, etc., is that it still represents a THEORETICAL win advancement.  MVP awards are based on ACTUAL performance which ACTUALLY led to actual runs and wins.  Probably runs and RBI are as close to any stats that you can use to evaluate for an MVP award.  Obviously in the VERY long run, both WPA and lwts will lead to the most runs and wins, but in one season, the chance of the player with the highest lwts or WPA actually causing his team to win the most number of games of any player, is VERY small.  The same goes for defense, actually, although we tend not to necessarily remember “clutch” defensive players, and we certainly have no recording of them.  However, for purposes of MVP awards, as with offense, the ONLY defense which counts is that which helped (not “theoretical” kind of help - but actual help, where the actual outcome was WINS) his team win games. Do you think that if a player had a high lwts or even WPA but he NEVER helped his team win any games (which is unlikely, but possible) he deserves consideration for MVP?  I think not.

So, while I am glad that at least Tango started with WPA (which, as I said, is better but not all that good for MVP evaluations), can we please NEVER utter the words lwts, WARP, OPS, VORP, or some other context-neutral stat in the same sentence as MVP?

Now, park is another thing.  Of course the park should be considered if we are going to look at any stats other than win shares.  What if a team played in a Little League park such that almost all of their players had a million runs, RBI, HR, their averages were .400+, etc.  Should they all warrant consideration for MVP?  Of course not.  The only thing that matters is a player’s contributions to his team winning games and/or making the post-season.  In a small park that is going to be with a batting avg., HR, RBI, runs, etc. that is greater than in a large park.  If the writers do not realize this, then they are not evaluating players for the award properly.

Every year we trot out the same guff—so let me trot out a subjective, non-statistical version.

The Phillies’ astonishing story also plays a role. If you define valuable as in value to the ballclub then there is little doubt that Rollins’ contribution was monumental (as was Howard’s and Utley’s).

Given the end of season excitement this race was going to go one of two ways: to the Rockies and Holliday or the Phillies (Utley/ Rollins/ Howard). All the other candidates were just that, candidates.

Had the Braves won the NL East then I suspect Chipper would have had a good shot at the award

MGL, how can you call WPA/LI context-neutral? It is the single-most context-specific measure out there!

The only difference between WPA/LI and WPA, is that WPA rewards the player for being in the right place at the right time (a run is not a run), while WPA/LI doesn’t do that (a run is a run).

But, WPA/LI is sensitive enough to change the value of the walk, the HR, the groundball that moves the runner over, etc, based on the game state.  WPA/LI says that Rollins was +3.1 wins.

A context-neutral measure (like LWTS) would say he’s something like +2.0 or +2.5 wins.

(All numbers not park adjusted.)

It’s very possible that Rollins made the most of his events.  That is, while the HR is worth 1.40 runs each, Rollins’ may have been worth 1.50, because he hit them when they counted more, etc, etc.  And remember, we are not including the leverage aspect of the game state.

WPA/LI is the context-specific wOBA.  That’s the great thing about WPA/LI: it’s context-specific, but doesn’t reward the player for the high leverage.

MGL,

You guys are on a whole different level than I am. Usually I’m content to just read and learn.

However, to keep this short: Wouldn’t you be rewarding a player for luck? The highest lwts or whatever may or may not have caused a team to win the most games, but couldn’t it still be argued that the player performed the best? I think I’d rather base an MVP award on performance than partly on fortuitous circumstances.

I can at least see it both ways.

MGL’s premise seems to be based on that value can only be earned if it directly contributes to a win (meaning that you participate in scoring or preventing a run, in a win).  This would imply that Vlad hitting a grand slam to tie a game in the playoffs or Endy making The Catch has no value in this case.  This would also imply that going 0-fer in a win has no negative value, since the team won, so what does it matter how he did.

The reader has to choose whether to accept the premise or not.  The process follows from the premise.

I on the other hand only believe in contributions in helping to try to win, which is the only thing a player can control.  And WPA, or WPA/LI, satisfies that.

Some people believe that winning a game needs lots of breaks, that the whole season is influenced heavily by luck. Therefore, their premise is that it doesn’t matter how a player actually performs in game situations, and doesn’t matter if the team wins or not.  Therefore, they will only look at the “seasonal” line (OBP, SLG, etc) of a player.

Finally, some people have no idea what their premise is, and will “consider everything”, which is a b.s. position as far as I’m concerned.  You have to start with your premise and follow the process to support that.

MGL, while I don’t support his view, at least is consistent in his view.  Most people have no idea what their position is, and will use the catchall “consider everything”, which really means they take a shot in the dark.

Just out of curiousity where does David Wright’s numbers stack up with the above.....i thought he definitely deserved more consideration.

OK, I understand those things. Mainly what I was asking was why MGL thought that was the one true way. Maybe that’s what you meant when you referred to him being consistent.

I guess my contention is that players can only control a limited number of things, so let’s reward performance for what they can control. Otherwise we could just take the five or 7 or 13 best players in a league and pick a name out of a hat for the award.

Not that I’m trying to change anyone’s opinion, I’m just curious if if my position is a reasonable one.

All three positions are reasonable.

The unreasonable one is the fourth one, the catchall.

Yes, you could do it based on “true talent level”.  After all, we know how great Pujols is.  What if he did everything he did last year, this year, but that this year, Tulo and Cameron et al all played fantastically against him.  See, we sometimes treat his actual performance as something that is his, when in fact, it’s based on the context.  And Pujols may have simply not caught the breaks.  What if opposing fielders muffed and threw balls away when Rollins was batting?  What if pitchers became Cormier around him?  That is, by pure luck, Rollins hit as well as he did.

So, you can simply say that “I’ll give the award to the best true talent player”, and that would be Pujols.

I do think that crediting Rollins with 2.0 for defence on top of 0.5 for positional adjustment in 2007 is really stretching the data.  The Fans Scouting Report (22 ballots) had Rollins as an 82, 2nd on the team and remembering that this is position-neutral evaluation, behind Victorino.  The FSR had Reyes at 83.  That is essentially consistent with a +5 shortstop, or in the vicinity of where Dewan had him. 

I would hope that Ripken would have been 90+ and Ozzie would have been 93-4 or so, with his only non-5 probably being his arm.

It doesn’t matter that he was 2nd on the team, unless you think the Phillies fans were being way biased. 

In any case, the conversion to runs is:
(Fans - avg) * 0.7 * avgBIP/4

For a SS, avg is around 58, avgBIP is 5.

So, a top SS, like Everett, would score as
(+30)*0.7*5/4= +26 runs

Fans had Rollins, Tulo, Reyes all at around +20 to +25 points, which is consistent to being around +20 runs.  It’s not stretching anything, if the Fans are right.  And, like I said, Dewan, from 2005-07, has Rollins a smidge away from being the #2 fielding SS in MLB.  So, the Fans see what Dewan sees.  Dewan, in 2007, didn’t see it in Rollins.  That doesn’t mean that it didn’t happen.

Dewan had Rollins at +14/yr. over those 3 years (and Pujols at +24/yr.). 

My point with respect to the fan evaluation is that over 22 ballots, the data is pretty unreliable and you can see that from the position-neutral evaluation of Victorino.  The most accurate of the fans ratings in my view were the Red Sox ratings, with the number of ballots probably being the key factor.

I suppose you can argue (not that I buy it) that Rollins had much more defensive value than Pujols; in that respect, I guess that this year’s vote isn’t exactly like Mauer/Morneau from last year…

The number of ballots doesn’t make it unreliable.  I’d get the same results with 44 or 200 Phillies fans.  When you get to 16 ballots, r=.90.  This is a statement of fact.  And researchers would kill to get something like that.  And I get it with 16 ballots.

The ballots will be very consistent.  They may be consistently biased (which is another issue, huge unto itself), but they are consistent, and getting more fans won’t increase their reliability.

Basically, the Fans’ evaluation of player’s fielding, given 16 ballots, is as *consistent* as having 1800 PA for a hitter.

Obviously, the bias potential is much lower for the hitter.

But the point is, more ballots won’t give you more consistency.  And more ballots won’t give you less bias.

***

What’s wrong with Victorino being considered the best RF in baseball?  I believe David Pinto’s PMR has him as #1.  Dewan has him as +13 outs made per 162 games.  If anything, it’s vindication for the Phillies fans that they got Victorino #1.

To argue that a player should get “credit” in the context of an MVP type award for things he does that increased his team’s probability of winning during the game, regardless of whether the team actually won or lost the game is outrageous, in my opinon.  That notion completely destroys the definition of the award!  Imagine a player (this is unlikely but it helps in arguments like this to look at an extreme) who racks up a sky high WPA or WPA/LI and yet his team loses every game.  You wouldn’t and you shouldn’t give him an eyelash of consideration for MVP!  It’s not his fault of course, and he did the best that one could possibly do relative to everyone else in the league, but that is too bad, so sad.  That is the way life works and that is the way an MVP award works.  It AIN’T the award for the “best player in the game” and it ain’t the award for the player who “gave his team the most theoretical win advancement.” It is an award that credits a player for contributing in some significant way to actual, real, cold, hard, cash in the barrel wins. The ONLY people in the world that don’t follow that line of thinking are SOME (certainly not me) sabermetric analysts.  An MVP award MUST be considered based upon subjective criteria AND stats that are commensurate with actual runs if not actual wins.

”...commensurate with actual runs if not actual wins”

That looks like a contradiction with your earlier bolded statement.

Also, here’s a situation: heroic player brings his team from 3 runs down in top of the ninth with a 2-out grand slam.  In bottom of inning, closer gets first two outs easily, then walks a batter, and clumsy right fielder drops an easy fly ball and knocks himself unconscious.  Two runs score, heroic player’s team loses.  Does heroic player still deserve no credit for this performance?

In the NHL, it has happened, several times, where the Conn Smythe trophy (playoff MVP) went to a player on the *losing* team.  This usually happens with a goalie (he stands on his head, but the team around him can’t help him).

I would say that there are many people who think of value in terms of contributions *toward* winning, regardless of the final outcome.

In team sports anyway.  It’d be real hard to argue that if Federer plays Nadal, and Nadal wins that Federer is more “valuable”.

For what it’s worth, I don’t care at all about the winner of MGL’s MVP award.  When MVP discussion season comes around, I use it as an excuse to discuss who the best player was in the past year.  I occasionally call that player the “MVP”, because that’s what everybody else is talking about.  I’d rather discuss something interesting/important than a useless (albeit “accurate") MVP definition.

Greg/20: the example I always use is a few years ago in the playoffs, Vlad hit a grand slam to tie the game against the Redsox in a game the Angels lost.  Or Endy’s The Catch.  MGL’s premise is that these games earn these players no MVP considerations.  They may as well have struck out 5 times with the bases loaded.

It’s based on the premise: do players have value in their contributions toward winning, or do they have value only if their team won.  Kinda like two companies bidding for a contract.  The losing company could have the best sales guy on the team, but the winning company had the better team.  Losing company gets no contract, and the top sales guy may as well have called in sick.  He gets no “MVP sales bonus”.  That’s where mgl is coming from.

I understand the premise, I just disagree with it. 

If that were truly the way most people thought, then for the last month we would have been reading about Rollins OPS+ in winning games vs Holliday’s OPS+ in losing games, and wining game/losing game splits for every other statistic that gets talked about, etc. etc.  You see that sort of trivia once in a while on a Fox broadcast, but otherwise people don’t talk about that split, because they don’t really think that way.  Point me to someone blogging about how many of Rollins SB’s came in winning games - I doubt such a thing exists.  Unless this is one of those things everyone thinks about but no one discusses

The what-if’s about hypothetical players whose team always loses are a bit silly, too, as 98% of MVP debates are between players on teams whose win % are very similar.  We’re not talking Andre Dawson here in 2007.  I’ll admit, I have a hard time supporting an MVP candidate from a last place team, but that’s because my initial guess would be that said player’s stats were hollow, else they would have led to more wins.  But, if someone showed me that actually that player contributed huge WPA, but was just surrounded with stiffs, I’d be much more sympathetic.

I will also grant that anecdotal recollection plays a huge part, and that heroic deeds that happen in winning games are more prominently recalled than those from losing efforts, but that’s not quite the same thing, IMO.  Maybe that’s what MGL is referring to…

[C]an we please NEVER utter the words lwts, WARP, OPS, VORP, or some other context-neutral stat in the same sentence as MVP?

MGL, please realize that many of us don’t agree with your premise in the least.  We believe that a player should be rewarded for what he does, and what he does alone.  If a player on a poor team hits a lot of solo HR because no one’s on base, and some other guy on a good team hits the same number of HR but with a bunch of guys on base, well, I simply don’t believe the second guy has created more value for his team.  Those guys on base aren’t his accomplishment, they aren’t his value.

Imagine a player (this is unlikely but it helps in arguments like this to look at an extreme) who racks up a sky high WPA or WPA/LI and yet his team loses every game.  You wouldn’t and you shouldn’t give him an eyelash of consideration for MVP!

Of course I can.  This player created substantial value to his team—it is not his fault that the rest of the team is poor.  The definition of the award, per the BBWAA, asks voters to consider the “[a]ctual value of a player to his team, that is, strength of offense and defense.” This player has provided that. 

The guidelines also state, “The MVP need not come from a division winner or other playoff qualifier,” which speaks for itself.

As Tango says, there are various ways of looking at the question, and you are welcome to your view, and you are consistent in it.  But many of us interpret it differently, and pursue the question along other lines.

The performance based on whether there are other men on base, I think, is critical in determining value.

Imagine other sports, like the QB, the playmaker in hockey or basketball, they make beautiful passes, picture perfect, but the receiver can’t hold on to the ball, or put his stick on the puck.  Do those things have value?  Does the value exist, because of what the player does, or does it exist because of what his teammate does with that?

You can consider everything a state, so that if Gretzky or Magic make the perfect pass (and the average player is there to receive it), then their team’s win expectancy has gone up by +.05 wins.  But if their teammate can’t hold on, the teammate get a -.07, and loss of possession, at which point, there was a turnover back to our hero’s team (+.02 for someone on L.A.). MGL, if I understand him, argues that it’s ridiculous to consider every little thing like this, since it hasn’t led to anything real.

The question is which discrete events to count, and what contexts to consider.

Maybe you want to give every great pass a +.05 win, regardless of whether the mate holds on to the pass, and/or scores on that play (which is Hawk’s argument). 

Maybe you only want to give a great pass +.10 if it leads to a score and a win (which is MGL’s argument). 

Maybe you want to give this particular great pass +.03 because the better option would have been to skate it in, since there was no one covering the passer to begin with (he dumped it too early).  This is my argument.

All hold ...uhm, value. I can accept a reasonable argument for either of these three scenarios.  What I can’t stand is people who say “everything”, which really means nothing.

why should fan evaluations be used to evaluate a player’s value to his team fielding?  assuming the fans are correct, the evaluations still have nothing to do with actual events that took place during the season. 

why is there so much discrepancy between dewan and uzr’s ratings of pujols’ fielding?

(if dewan is right, i probably would’ve voted for albert)

I don’t think the voters agree with MGL’s premise.  If they did they could not have voted A-Rod an MVP award his last year in Texas.

Matt/27: sure they do.  The Fans “see” something.  They saw the collapse of Betancourt’s throwing for example.

The correlation of Fans to PMR is as high as UZR to PMR.  I doubt anyone here would have believed that would be possible, but there you have it.

Think about that.  Two PBP-based systems correlate to each other at around the same level to a completely subjective independent method.

Tango/29, my take on Matt’s post (27) is that he is saying that they are essentially a context neutral evaluation (the fans scouting report, that is). The fans (I don’t think) aren’t thinking about contributing to winning as in base configuration, what the score is, timing, etc. The fans scouting report is like what VORP, or the other context neutral offensive stats are to offense. I suppose you would have to somehow incorporate WPA into a fielding stat to get something that would be useful for the MVP, depending on your definition of MVP.

Anyhow, great thread. I had never really thought about it quite like this, but it definitely doesn’t make sense to use VORP, OPS, etc (again, depending on your definition). Tell you the truth, I never really think about these awards too much, other than “the voters don’t get it.” Maybe that’s not exactly right.

The voters, fans, etc. do NOT remember when a player hits a grand slam to go ahead in the 9th and then the team loses the game.  That is the ONLY reason why the player gets credit for that kind of performance. The team wins X amount of games, the player had a great year, and the voters and fans assume that the player contributed mightily to those wins.  If they knew that he didn’t, he would get very little credit and deservedly so (for puspodes of the award).

Rodriguez wins it with Texas because Texas won X amount of games (70, 80?) and it was assumed that A-Rod contributed mightily to those wins, which he probably did. Had Texas won NO games, I guarantee you A-Rod would have won NO awards, even with exactly the same great performance.  That is my point.

After the fact, a player CANNOT be VALUABLE to a team if he does not create REAL wins.  An employee in my company, AFTER THE FACT, could not have been valuable if my company earned no money (although you can make an argument that the company would have lost even more without the employee so it is a bad example).  Doesn’t mean the employee didn’t do a great job and I don’t want to give him a raise and keep him in my employ.  Just means that the company lost money through no fault of his own.  Same with ballplayers.  NO VOTER OR FAN is going to give a player credit for MVP unless they think that he contributes to cold, hard wins.  I don’t think you can dispute that.

SKY, what does that mean, “Best player?”

"After the fact, a player CANNOT be VALUABLE to a team if he does not create REAL wins.”

But this begs the question:  how many wins did player X create?  And despite having dozens of metrics, we don’t actually have an answer for that. 

Take this sequence:  single-2R HR.  Post-hoc, how many runs did each player contribute?  If the first player had made outs, that’s one less run, so maybe he gets 1/2 the credit. But if the 2nd player had made out, that’s a 2-run swing, so maybe he deserves 2/3 credit.  Linear weights would give us an entirely different answer.  WPA gives us an answer, but it’s not post-hoc, so gives weight to the current score that isn’t relevant at the end of the day. (It also doesn’t credit the first hitter with good timing in getting on base in front of the HR).  It tells us how much the player improved it’s probability of winning each time he hit, but not how many games he truly won for his team. 

Then there’s the game outcome. If the team wins 2-0, how much credit should the hitters get vs. the pitchers and fielders?

I don’t know how you can answer these questions without employing some kinds of context-neutral tools.

In reference to 21 - the NHL isn’t the only league that gives out MVP awards to losing teams. Mike Scott won the NLCS MVP against the Mets in 1986.

I prefer WPA as the main offensive consideration for MVP talks.
Another issue is how much consideration the team’s final place in the standings should receive. I assume MGL’s only going to consider teams that made the playoffs because he seems to be taking an all-or-nothing approach.
I think context is important, but I don’t go that far. Hanley Ramirez’s performance has to be considered less valuable because his team was out of the race early in the season, so the games were somewhat meaningless. On the other hand, David Wright (who would be pretty close to Rollins on the chart above if he was included)’s team was in the race until the final day off the season, so all of his positive performance came in meaningful games.

Only teams that make the playoffs?  Not at all, unless it says that in the “instructions” or guidelines, or it is obvious.  I don’t think either is the case.  Winning teams or losing teams?  Doesn’t make that much difference to me, although with playoffs and losing records, we get into subjective/grey areas, which is also well and good.  An MVP-type award has to have a large subjective component.  There is no way around it.

All I ask is that we look at a player’s contribution to real wins - that’s all.  If it’s A-Rod contributing to 10 extra wins for a losing team like Texas, fine.  If it is someone else contributing to 6 extra wins for a playoff team and you want to give that player extra credit for propeling his team into the playoffs, fine.  Just don’t tell me that so-and-so deserves consideration for MVP, regardless of how well or poorly his team did, and no matter how well the player did, when he DID NOT contribute to many or any actual wins.  That’s all I am saying.

Just give a player credit for contribution to actual wins.  There are many statistical ways to do that and none of them are going to be foolproof.  But…

“Contribution” HAS to be contribution to actual wins and not theoretical ones otherwise it can lead you down the path to an absurd result (reductio absurdum it is called in legal analysis) - that of a player winning an MVP award when his team wins zero games.  If you want to give ANY credit for theoretical win advancement, be it WPA, WPA/LI, or lwts, you HAVE to be wiling to live with that result, no?

MGL is referring to wins (as in games), not wins in series.

So, if Mike Scott gets three shutouts, but the Mets win a 7-game series, Scott could get the MVP.

In the NHL, they have the “3 stars of the game”, and almost always, they give 2 stars to the winning team, and 1 to the losing team.  That’s the way I see it: contributions to trying to win.  In MGL’s scenario, if you were to give 10 MVPs per game, all of them would come from the winning team.

I admit I am not as well versed in sabermetrics, win shares, and the new advances in statistical measurements in baseball as I should be, but I have been a fan of this game for nearly 30 years, and it is when I see a premise like this that I disagree with the idea value can be measured soley by numbers.

“MGL, please realize that many of us don’t agree with your premise in the least.  We believe that a player should be rewarded for what he does, and what he does alone.  If a player on a poor team hits a lot of solo HR because no one’s on base, and some other guy on a good team hits the same number of HR but with a bunch of guys on base, well, I simply don’t believe the second guy has created more value for his team.  Those guys on base aren’t his accomplishment, they aren’t his value.”

To me, the premise that all at bats are of the same value is faulty.  A hitter coming up to the plate with men on is in a more valuable situation to his team than if he came up with no one on.  Yes, he was not directly responsible for the baserunners ahead of him getting on. But if he does get a hit, he is responsible for the runners in or advanced for the next hitter. 

So in LA Waterloo’s example, I patently disagree. The second guy did create more value for his team. I don’t believe you can assume that both at bats (one with none on, one with men on) can be measured the same way.  The pitcher doesn’t pitch the same way in those situations.  The batters approach to the at bat is different as well.

I would love some more education on this, so fire away.

MGL/34

So, which stats have you examined for the 2007 season that are split out by wins/losses (i.e. someone’s performances in wins vs. performances in losses)?  I haven’t seen any is why I’m asking.  But maybe I haven’t looked hard enough.

Sorry, that last post came out a little harsher than I intended, I am not usually the snarky type. 

But seriously, I haven’t read anything from anyone who is looking at it that way (crediting only performances in wins), and I was surprised at your (MGL’s) vehemence on the point.

Haven’t you got some “give” in your viewpoint?  If a salesman works really hard on an account and does some really inventive work, but still gets beat out by a superior competitor, don’t the people around him/her see that and give some credit to the performance?  In my 10 years in the private sector, they have - comes time for promotions, those accounts count on the plus side, and lackluster performances on accounts that were won are noticed and judged accordingly, too.

Greg, your post was not snarky at all, not that I care.  I only care about content, not “tone of voice.” The only people that USUALLY care and complain about tone of voice are the ones who lack anything substantive to say so they resort to attacking the person whose content they disagree with or don’t understand rather than respond to that content.

Anyway, I have not given it enough thought, and frankly have no interest in MVP type awards, to come up with any objective methods of evaluating players for MVP.  As I said, runs and RBI are a good place to start.  They don’t necessarily lead to wins of course, but they are more correlated to runs and wins than any sabermetric stat.

And yes, of course I have some give in this area, because I concede that the entire area of evaluating MVP’s is grey.

Nick the Greek, despite being racist , is right on the money.  Waterloo’s argument is about what a player has control over.  That is fine for determining who is the “better” player, who you would rather have on your team in the future, for projections, etc.  It has little relevance for MVP type awards.  NO ONE (except one-track sabermetricians) cares about context neutral or theoretical value after the fact in the context of giving out an MVP type award.  And well they shouldn’t.  Yeah, we all know that it likely wasn’t a player’s fault or his intention to hit a 3-run home run or solo home run or a game winning home run in the 9th or a game tying HR in the 2nd and his team eventually loses, or a grand slam when his team is down by 10, yada, yada, yada.  That is the mantra of the sabermetrician.  But that has nothing to do with MVP awards.  They are all quite different in the context of those awards and well they should be.  He who helps his team ACTUALLY win games and pennants and post-season berths and World Series, gets the trophy and deserves it.

Here is the deal in a nutshell:

One player hits 2 solo home runs early in a game.  His team gets blown out.  Another player hits a grand slam in the bottom of the 9th, his team down 3 runs, after hitting a GS HR in the 8th to bring his team from down 7 to down 3.  Both players have the same lwts.  Who do you think will get the most MVP credit in the eyes of virtually anyone but a percentage of stat geeks?  That is why lwts or any other context neutral stat is worthless, although obviously lwts will tend to correlate with actually winning games.  A.

Player A has a WPA of .2 in one game.  They end up being down by 3 in the 7th inning and lose the game by 6.  No one will even remember this guy and rightfully so.  If you try and give him an MVP award for this game, you will be laughed off of your computer.  Player B makes a great catch early in the game with the bases loaded and 2 outs, and then hits a 3 run HR in the 7th with his team down by 3.  Let’s say that he gets .2 in WPA for this game (I don’t know if that is possible, but let’s assume that he lost a little WPA in his other 4 PA).  Both players have the same WPA for the game.  Would anyone be surprsed if player B wins MVP of the game?  That is why WPA is better than lwts, but it still stinks.  NO PLAYER is going to win an MVP without contributing to actual wins. Period. If you try and give an MVP award to someone who contributed theoretical wins, either through lwts, WPA, or WPA/LI, but little or no actual wins, you will again be laughed out of the room. The reason that some people on this thread think that WPA or even lwts is OK is that often the guys with the high WPA or lwts ARE going to contribute to a lot of wins.

A guy with a high BA is likely to have a high lwts too.  If we used BA to determine who the best players are, these same guys would scream bloody murder, and justifiably so.  That is why we don’t do it.

I have not read the article, but this is on the right track:

http://www.baseballdigestdaily.com/bullpen/index.php?option=com_content&task=view&id=358&Itemid=39

Looking at the actual MVP guidelines, it seems to me that there are 3 relevant passages:
1)"Actual value of a player to his team, that is, strength of offense and defense.”
2) The MVP need not come from a division winner or playoff qualifier.”
3)"There is no clear-cut definition of what Most Valuable means. It is up to the individual voters to decide...”

MGL is operating from an assumption that his definition of value is the correct one. But in this context, there ISN’T a ‘correct’ definition. There is nothing written there that suggests that looking at performance in losses is improper, or that looking at ‘context-neutral’ performance is wrong.

Too much emphasis is put on the word ‘valuable’ and what it means, as though that term was ordained by God. But maybe they spent only only a minute selecting the term ‘valuable’, and might have, on a different day, settled on something else, like “best performer”, or “most productive player” (which, of course, are themselves somewhat ambiguous).

I, for one, prefer the haziness of the actual guidelines, to an unwinnable attempt at perfect clarification.

#41, I completely agree that too much emphasis is put on the “valuable” part of the acronym.

I don’t think it does stipulate, but if it was clear to me that MGL’s take was the only correct interpretation, then I simply wouldn’t care about it.

I don’t mind interpreting it as performance. As in, the player that would be most valuable no matter where you dropped him in the league. No matter how “lucky” he got with leverage situations and base-out states, how many games his team won, etc, etc. “Most valuable to a team in 2007” instead of “most valuable to his team in 2007,” maybe. Eh, that didn’t come out great, but whatever.

If I were convinced that we needed to base the MVP on Runs Produced or whatever, then I’m just not going to care who “wins” that. It’s of no use to me. Unless the winner gave a “I couldn’t win this without my teammate’s speech” then I’d chuckle.

Perhaps people are just saying that they care about the same thing, and most of this is basically semantics. I doubt anyone voting for the MVP is pulling out slwts or WPA anyway. I might as well care about what I care about, and look at that.

Don’t people find fault in the idea of “Context neutral”.  Yes you might be able to statistically adjust for ballparks being different, or position played on the field.  But you can’t truly have a neutral comparison, unless all batters face the same pitchers the same number of at bats and the same number of pitches per at bat, etc. 

I guess the result of the adjustments make the numbers more comparable, but I wouldn’t call them neutral.

I guess I am arguing how the voters and 99.9% of the fans would interpret “valuable” which may be a little tautological.  But if you read the last couple of paragraphs in my post #43, it is clear that I am arguing from the point of view of those fans and voters who will not EVER vote for a player based on theoretical contribution to wins or win percentage, other than to the extent of not remembering each individual performance and whether it actually contributed to wins (IOW, if a team wins a lot of games and/or makes the post-season, and a player has a great “neutral” performance, it is assumed that he must have contributed a lot to those wins, whether he actually did or didn’t).

MGL/44

I agree completely with that last part, I believe that is what most people do.  And it’s not an unreasonable thing to do, IMO, because for most real player seasons (as opposed to extreme hypotheticals we use to make a point), that assumption will be true ...

The key to getting the best MVP selections is not to dissect the term ‘valuable’ to its most logical endpoint, and then select the stat which comes closest to that.

The key is to give some guidelines, or boundary points, and then let the voters have the freedom to operate within those guidelines. IOW, the key is to find the best qualified voters.

MGL deserves to be considered for that group, but so does Tango, B James. E Weaver, J Shuerholz, and maybe even a smart active player like G Maddux.

The real problem with the MVP framework is the BBWAA presence/dominance.

http://baseballanalysts.com/archives/2008/09/in_the_next_mon.php