THE BOOK--Playing The Percentages In Baseball

The underrepresentation of pitchers is probably intentional, and in my opinion, wrong.  In his top 50, he includes Johan, Webb, Sabathia, Peavy, Haren, Beckett, and Verlander.  Seven pitchers against 43 position players? That’s not anything close to balancing risk and reward.

Since pitchers get 40% of the salaries, you’d expect some 20 pitchers.  7 does seem pretty low.

The only way to know is to do a WAR aging curve similar to what I did with Win Shares.  (WS terribly underweights pitchers, so maybe Rob got fooled by that?)

Just one word....Melky???????

Melky is a good example.  A league-average hitter, a position-average fielder (according to Fans/UZR… Dewan hates him), but playing a premium position.  That makes him +0.5 WAA (wins above average).  He’s in the AL, so he’s a +3.0 WAR (wins above replacement) per 162.  Give him 80% of that, and he’s +2.4 WAR.

His Win Shares are comparable, at 13 WS on average.  (The average nonpitcher if given 80% of the playing time should get: 81*3*.65/9*.80 = 14 win shares.)

So, the question on the table is: how much is a league average player worth, if he’s only 23 years old?

Breaking out our trusty 5-yr formula:
23 + 0.64 * 14 - 0.78 * 23 = 14 per year x 5 years = 70 win shares.

Basically, we should expect a league average player, at age 23, over the next 5 years, to be an overall average player.  This means that his likely improvement is balanced against potential attrition (injury and whatnot).

This is pretty valuable to know.  We can break out our salary chart, and give no aging to a league average player (2 WAR), and we’d need to pay this guy 56MM over the next 5 years.

Does that sound like the 46th best player?  I dunno.  Probably not.  You’d have to work out the rest of the group to see.

Of course, if Cabrera is really +2.4 WAR (as opposed to the league average +2.0 WAR), that gives him a bump.  That’d be the equivalent of say 15 win shares instead of 14.  That gives him a slight bump to 73 win shares over 5 years, meaning a slight twinge of aging.

Really, all this list goes to show, is that you need to know two important things: how old is he, and how good is he now.  Those two things are hard to figure out how to balance, unless you have a trusty formula like above.

"Since pitchers get 40% of the salaries, you’d expect some 20 pitchers”

I’m not sure that one follows the other.  Given the injury risk for pitchers, I suspect that over any given 5-year period, fewer than 40% of the top 50 players would be pitchers.  Just doing a very crude estimate of RAR over past 5 years, with no defensive value included for position players, I get 11 pitchers out of the top 50.  Others here could do a much more precise estimate, but it wouldn’t surprise me if far fewer than 20 of the 50 most valuable 2003-2007 players were pitchers. 

So if one’s goal were to pick as many of the actual top 50 as you can, it may be right to pick mostly position players and a few stud pitchers.

I like Melky, and with his youth I think he’ll be a good player, but he’s not a league average hitter yet.  His OPS+ is 90, he’s been a below average hitter each of his two seasons, with 2007 being worse than 2006.

Without creating my own list, it’s real hard to see him as top 50.

Melky is rated as a below average defender in center according to Dewan’s +/-, PMR, and Justin’s fielding translations. I’m not sure how UZR sees him but I’m also not sure how UZR sees Ichiro as the worst CFer in the AL.

I said that UZR has him as average.

The number of systems that see him as otherwise doesn’t give less weight to UZR.  UZR should get roughly 30% of the weight.  Pinto 25%, Dewan 25% and Fans 20%.  And the larger the sample, the more I’d give UZR and less I’d give Fans.

***

Rally: right, I spoke too fast about the 40%.  If I look at say Fantasy picks for last year, I had 50 nonpitchers and 11 pitchers, at 20$ or above.

Clearly a crude measure, but it certainly gives an initial support that at the top end, getting say 20% of the players as pitchers might be fine.

Ideally, you would do this as WAR, along with the 5-yr aging curve for WAR.  So, perhaps 7 pitchers is not so out of line.

I’m thinking that, except in cases where there’s a disproportionate strength of pitchers or position players, the split should be more in the 38-12 to 36-14 range. 

Just eyeballing it, I think you’d need to make room for Erik Bedard, Cole Hamels, Felix Hernandez, and Fausto Carmona, and then probably one or two from the Aaron Harang/Scott Kazmir/Chad Billingsley/Dustin McGowan/Chris Young tier of pitchers. 

I don’t think a strong case can be made for Asdrubal Cabrera, Justin Morneau, Melky Cabrera, or Dan Uggla actually belonging on this list, so you toss them out the window and replace them with the first four pitchers above.  Then you can fight about a couple of the pitchers from the second tier against the marginal picks like Peralta, Pence, Kemp, Gordon, Gonzalez, and Howard. 

But 43-7? No way.

There are two separate questions here:

1) how many of the 50 will actually be pitchers?  I’m not sure, but 12-14 seems reasonable.

2) Which 50 players have highest probability of ending up in the top 50?  On this question, it may be reasonable to have fewer pitchers because the uncertainty is higher.  There are a number of young pitchers who could end up on the list, and a couple of them will, but perhaps the odds for any one of them may be less than for Kemp or Gordon. 

The upside for Felix or Hamels may be top 20, but would we truly be surprised if in 2013 they weren’t on the list?  No.  But if David Wright isn’t on the list, we will be surprised.

Let’s try to be tough on pitchers, and give them a tougher aging pattern than hitters.

For a 22-23 year old nonpitcher, let’s presume that over the next 5 years, he’ll be the same quality, reasoning that how much he improves will be balanced out against injuries.  We’re being a bit generous to nonpitchers, as we saw from the Win Shares aging curve.

So, a 3 WAR nonpitcher aged 22-23 will get himself 15 WAR over the next 5 years.

A pitcher though, will suffer more.  Let’s knock out both quality (drop his win % by .005 each year) and quantity (drop 5% from his IP).  We start him at .550, and 189 IP (21 full games).  With repl level of .380, we get this for a pitcher:
win% G WAR
0.550 21.0 3.6
0.545 20.0 3.3
0.540 19.0 3.0
0.535 18.0 2.8
0.530 17.1 2.6

His total WAR over 5 years is 15.3, which is similar to the nonpitcher.

Clearly, Melky is not a 3 WAR player, and clearly Felix is at least a .550 pitcher, with at least 21 G in his bag.

And, I was being generous to nonpitchers and a bit harsh on the young pitchers.

It certainly looks like Rob’s list is biased against pitchers.

I hope, just hope, that Rob didn’t rely on Win Shares to make his list.  Win Shares is horribly biased against starting pitchers.

Tango:  are those IP estimates based on real pitcher performance?  i.e. Do we expect a good 23-yr-old pitcher to actually average 856+ IP over the next 5 years?  Seems a bit high to me.  Only 24 pitchers who were age 26 or younger in 2003 have thrown 856 IP over the past 5 seasons.

And even if that’s right, what we want to know is not the expected values for Felix and Melky, but the probability each will have 20 WAR over the next 5 years (or whatever it takes to crack the top 50)—not quite the same thing. Maybe I’m giving too much weight to the Kerry Wood/Rick Ankiel/Alan Benes factor—all of whom might have seemed like locks for this list at one time—but I have to think the probability of Felix putting up just 5 WAR over the next 5 seasons is much higher than for Melky (though his upside is obviously higher).

All that said, I totally agree with your point on WS and starters.

Guy,

All just guessing on my part.

The point though is that to answer the question, you need to come up with a reasonable model. That was my attempt.

I’m looking for others to do the same…

FWIW, UZR projection for Melky in CF is +2.  His batting projection is -5 (compared to AL average at all positions), and his total Slwts projection this year is around 2 WAR.

Terry, yes, UZR did not like Ichiro in RF and hates him in CF (at least it is consistent across positions).  What that means, I have no idea, other than it means that the UZR methodology does not “cheat.” By that, I mean that if it were absolutely perfect, it would be guaranteed to be wrong by a certain amount, for a certain amount of players, as long as the sample size it is working with is less than infinite.  I hope that makes sense to you and to all others who point out the various ratings for the various metrics that appear to be “wrong.”

I plugged the whole list into a spreadsheet, and added Win Shares for the last three years. Then, I averaged them (if the guy only had two full years, I did a two-year average, and if only one, I used 75% of that one year as the “average"), cross-referenced them to Tom’s aging patterns for Win Shares, and took a look. I ignored some guys, like Evan Longoria, who either lacked data or lacked enough data.

According to this, the projected leader is Pujols, at 125.35. Miguel Cabrera is a close second at 124, Wright is a close third at 123.77, and there are three others are over 100: Ramírez, 108, Sizemore, 106.65, and Mauer, 101.33.

Another observtion—Neyer’s list is NOT based on Win Shares and some formula. How am I so certain? Hanley Ramírez was ranked second, age 24, and his Win Shares for the last three years go 0-24-29. Right behind him is Miguel Cabrera, who is the same age and whose last three years of Win Shares go 29-34-30. I’m sure Rob looked at Win Shares, but he couldn’t have relied on them and ranked Ramírez above Cabrera.

Good stuff Charles.