THE BOOK--Playing The Percentages In Baseball

WinShares seems pretty outdated and inaccurate.  Why should I pay any attention to a ranking based on it, when we have WAR now?

Using WAR makes a huge difference for pitchers.  I tried repeating Pos’ experiment with rWAR instead of WS, and I get the following as 5 year leaders over that time period:

Joe Morgan, Mike Schmidt, Rickey Henderson, Wade Boggs, Roger Clemens (in 1990 and 1991), Barry Bonds, and Albert Pujols.

So Clemens goes from never being in the discussion to leading 2 years.  Tom Seaver, Gaylord Perry, Phil Niekro, Maddux, Pedro, and Clemens again were all 2nd at least once.  There were more pitchers in the discussion as well.

It looks like Larry Granillo, a commenter at Joe’s site, did something similar with rWAR at wezen-ball:

http://wezen-ball.com/2010-articles/january/revisiting-bill-james-qbaseballs-best-playerq.html

He has pitchers better represented there, too.

I don’t think that WAR should become the go-to stat for every need. It does a good job measuring relative value and is especially helpful for putting a payroll together.

But for looking back at history (the games actually played), I prefer the Win Shares philosophy. There are fewer hypotheticals involved. Win Shares focuses on the tangibles of runs produced and runs prevented. If I want to know the actual contributions a player made on the field, Win Shares seems better designed to answer those questions.

A brief example. In a game a slugging first baseman hits a HR and 4 RBI’s. It is irrelevant whether another slugger on the bench that day was equally capable of that output. The slugger who actually played was the one who contributed to the win. His contribution was more directly responsible for creating runs than was the shortstop who had one single in four at-bats, even if there was no comparable replacement on the bench that day. The shortstop’s value was in its rarity (calculated by WAR with the positional adjustment), but the slugger’s value was in the actual production of runs that contributed to the team’s win that day (which WS attempts to measure).

The Win Shares philosophy (even if the WS formula needs tinkering) still has a place, especially in HOF discussions and top 100 rankings (in other words, for past-oriented rather than present or future-oriented issues). It matches more closely with the games we actually witnessed, rather than games that might have been.

pac/3: you are horribly misinformed.

I don’t even know where to start with your post.  You need a two-hour refresher as to how Win Shares works, and I HIGHLY suggest you read the Rob Wood v Tangotiger debate here:
http://www.tangotiger.net/winshares.pdf

You can also read more Win Shares articles here:
http://www.tangotiger.net/#Winshares

And don’t tell me you had read all that already.  You will leave me exasperated.

Tangotiger - Fair enough, I’ll take a closer look at those articles when I have some time to give them a careful reading. The WS philosophy as I understand it is appealing, but it looks like those articles will give me some more to think about.

Bill James recently posted:

I don’t comment on other people’s analytical methods, essentially because all that would accomplish would be to

1) start personal arguments, and

2) expose my own ignorance. 

I get five questions a week here asking me to comment on this method or that one.  I don’t mean to be disrespectful of other’s peoples methods by not commenting on them; I’m just trying to give them their space.  The measurements I like best are my own; I assume this is true of everybody. 

There is nothing to be gained by picking at the things I disagree with about other people’s approaches.

(Bolding is mine.)

I’ve critically assessed hundreds of methods, and the only ones that devolved into something personal are those people who make it personal.  I can only presume that Bill has had the bad luck to be involved in debates with lots of people who don’t know how to argue on the merits, and instead rely on the personal.  It’s unfortunate that Bill’s personal experiences keep potential healthy debates from happening.

Number two is very big of him to admit.  Not many people admit their ignorance, and it’s better to be honest than to try to fool everyone else.  You’d be exposed at some point anyway.

As for the part I bolded, what would be gained is to not have people buy into something where there’s great evidence that Bill is so wrong on starting pitchers within the Win Shares framework.

What Win Shares is implying is that if you knew AFTER THE FACT how Pedro would have pitched 1999-2004, then he would NOT be in the discussion of greatest player of that time period, and (presumably) he could not support a salary that would pay him as if he were in the discussion of the greatest player of that time period.

Here for example are the totals from 1999-2004, using Studes’ win shares database from 2007:
246 bondsba01
193 rodrial01
177 sheffga01
175 giambja01
171 ramirma02
170 abreubo01
166 heltoto01
165 gilesbr02
164 jonesch06
163 thomeji01
161 sosasa01
159 kentje01
159 bagweje01
156 guerrvl01
155 delgaca01
154 jeterde01
154 tejadmi01
153 rolensc01
152 gonzalu01
151 edmonji01
150 jonesan01
146 greensh01
142 willibe02
142 pujolal01
138 johnsra05 <--
138 alfoned01
133 kleskry01
133 posadjo01
130 berkmla01
130 beltrca01
129 palmera01
128 boonebr01
127 garcino01
127 damonjo01
126 martipe02 <--
125 camermi01
124 ordonma01
123 alomaro01
122 rodriiv01
121 olerujo01

RJ is #25, and Pedro is #35.  Two pitchers out of 40 players.

Contrast this with the top 10 from Rally’s WAR DB:
WAR retroid
60.1 bondb001
45.8 rodra001
41.4 johnr005 <--
41.0 martp001 <--
38.8 heltt001
36.9 edmoj001
36.7 giamj001
34.9 roles001
34.5 jonea002
34.1 schic002 <--

Now, I ask you, in the “discussion” for greatest players of 1999-2004, after we get through Bonds, aren’t RJ and Pedro right there?

I should also disclose that WAR is a framework developed in this blog, and Rally implemented his version of that framework.