Win Shares (Merged Threads)

[nyy26wc]

I got Win Shares this afternoon, by downloading a digital copy from STATS's website.

I'm up to page 80 and am completely unimpressed.

The new version of the Historical Baseball Abstract was unimpressive, but at least had its occassional good tidbits, occassional good essays and occassional examples of the great writing and reasoning that James made his reputation on.

But, the rest was nothing more than a formerly great, yet past his prime writer living off his reputation.

But, in this book, the next thing I see that doesn't make me say that will be the first such thing.

[sweaver]

For those interested, the paper copy is $29.95, the downloadable version is $19.95, at the website www.stats.com There are assorted preview pages available in pdf format to help you decide.

[gyb13]

What do you think of his methodology in devising win shares?

[Skip]

I've been interested in the Win Shares book since the NHBA came out, but made a emotional decision that I'd refuse to buy it since my take on NHBA was that it was largely a big teaser for the WS book. I dont like being played like that. Not that NHBA was all bad, but .... urghhh.

The <expletive deleted> part of the whole thing is that I'm sure I'll eventually buy the thing anyway. Yeah, and I cry at the end of Field of Dreams every damn time too.

[Fritz Buelow]

Give in Skip - we can't help ourselves. I always say "I can't fit one more book in the house" - and end up buying them a few days later anyway. Baseball Junkies. We're hopeless.

[nyy26wc]

Originally posted by gyb13
What do you think of his methodology in devising win shares?

Through 80 pages, he's made no effort to explain why his method is valid.

In the beginning, he made his grand statements that we finally have a great way to assign a value to every season for every player in history and that will allow us to answer all sort of questions. But, while he's told us it's great, he's very weak on establishing validity.

He's explaining how you go from taking a player's stats and determining his wins shares from that stat line. But, he's avoiding the why should we give any value to that result question.

Put another way, so far the book has just a discussion of the mathematics, without any effort to make a convincing case that there is validity to the results.

[nyy26wc]

Originally posted by Skip
I've been interested in the Win Shares book since the NHBA came out, but made a emotional decision that I'd refuse to buy it since my take on NHBA was that it was largely a big teaser for the WS book. I dont like being played like that. Not that NHBA was all bad, but .... urghhh.

The James of the 1980s built his reputation by making a compelling case for why we should a method and then giving us the results.

When he failed to even make a valid attempt to convince us of Win Shares's validity in the NHBA, I suspected that the stat wasn't going to be as good as he had billed it. I suspected that, if there was credibility to the stat, Win Shares would have come out first, as the teaser for the NHBA.

[Skip]

Exactly! I couldnt get past thinking if James was "the man" that the methodology would have superceded the rest. I enjoyed the NHBA, but rank it with Neyer's Dynasties book in the interesting but not seminal category.

[Skip]

That being said, I should be clear that I dont yet have a copy of the WS book and havent downloaded whatever is available. But in truth, I am not expecting much - sorry to say.

[Fritz Buelow]

Lee - so, we won't be seeing WS on the next ed. of the SBE? ;)

[Ytown Tribe fan]

James gives his reasoning, his methodology, and a summary of his modeling in the New Historical Abstract. It's all right there. The new book merely goes into greater detail and has more players rated.

James has devised a way to measure the impact of runs created (specifically, marginal runs created) on wins, ERA+ and innings pitched on marginal runs prevented and, most importantly, individual defensive runs saved based on team defensive runs saved -- the single biggest breakthrough in determining defensive value ever.

IF you already believe in the validity of Runs Created, as opposed to runs and RBI;

IF you already believe in the Pythagorean method of determining W-L%;

IF you already believe that a pitcher's ERA+ is the single best measure of his ability, and not his W-L record or raw ERA, which are obviously affected by run support and the era and parks in which he pitched;

THEN you will find Win Shares to be a valuable method of determining the overall value of any player from any era, and perhaps the very best method for determining defensive value ever created.

Otherwise, pass. It's a matter of faith.

[gyb13]

Originally posted by Ytown Tribe fan
James gives his reasoning, his methodology, and a summary of his modeling in the New Historical Abstract. It's all right there. The new book merely goes into greater detail and has more players rated..
I'll have to go back to my NHBA, but I don't remember his explanation about win shares to be satisfactory. I may recall incorrectly, but he did some explaining, then left the door open for going into further detail in the WS book. I'll have to double check when I get back to DC...

[sweaver]

No more comments on this? It is, potentially, the most important book on baseball since, well, the original Historical Baseball Abstract. Anyone else download it yet? My cash flow is not yet sufficient.

[nyy26wc]

Originally posted by Ytown Tribe fan

IF ...

IF ...

IF ...



A few more IFs--

IF you believe that a player should be compared to "replacement level" instead of average.

IF you believe that James all of a sudden found the correct way to handle an apparently arbitrary defense system that he tried using in the early 1980s and just discarded.

Do you remember his defensive system from the early '80s, before he lost interest in it? He had a 40 point stat, a 30 point stat, etc., for each position and then assigned points based on the player's defensive stats in each particular stat.

He's revived the same idea, giving no strong argument for how he got right this time.

IF you believe people when they tell you their system works, without them making a strong effort to prove it

I've gotten up to page 100 and don't know when I'll waste my time finishing it.

[Fritz Buelow]

Originally posted by sweaver
No more comments on this? It is, potentially, the most important book on baseball since, well, the original Historical Baseball Abstract. Anyone else download it yet? My cash flow is not yet sufficient.

I ordered the hardcopy - still waiting on it. Can't comment until then. :(

[wyvern37]

I think there are two reasons why Bill James does not try to prove the validity of his system: 1) Win Shares is based on metrics that are not new. Hitter are rated based on Runs Created, Pitchers are rated on ERA+ and Fielders are rated on a system similar to Zone Rating. These measures are all widely accepted in the Sabre community. AND 2) He doesn't feel he needs to. I think that Bill James has reached a point in his career where he has decided that it is impossible to compare players from different eras. Therefore, he has developed a system that measures a player in terms of value to their own team, i.e. Win Shares.

The problem I have with the book right now is that I do not agree that a team that Wins 120 games is 20% better than a team that wins 100. In fact, I think that it is pretty apparent that this is not true. The competition faced by teams (even in the same division) can differ vastly. For example, some clubs (like Atlanta) may face nothing but top pitchers. Other clubs (say the Marlins) may have the luxury of facing a lot of #4 and 5 starters. These matchups may affect the won-lost records of both clubs. But does this change the inherent value of either franchise? No.

Nevertheless, I think Win Shares is a vast improvement over Total Baseball's rating system. The defensive ratings in Win Shares are much more sensible than anything I have ever seen before. They place a reasonable emphasis on defense without over-rating it, IMO. Defense pushes Honus Wagner ahead of Babe Ruth for the best single season of all time (1908). However, it does not prevent Ty Cobb from being more valuable than Tris Speaker. Both results make sense to me.

I think that a lot of people are going to dismiss Win Shares prematurely. I don't think they should.

Karim

[Fritz Buelow]

Karim, I take it you've read it, no? E- or paper?

[wyvern37]

Steve,

I got the E-book version because I was too busy to go to the store. I have skimmed through all of it and am about halfway through reading it more thoroughly. I have to admit that I long for a paper version. I probably will go out and buy the hardcover this weekend. It is just too annoying to read it onscreen.

Karim

[Fritz Buelow]

Thanks - only reason I ask is that STATS INC has replaced my order for the Baseball Scoreboard (not printed this year) with Bill's WS book. I thought it was supposed to come in March. Nada yet. :(

[Fritz Buelow]

Well, WS arrived in the mail today.
727 pages, huh?
Will let you know when I finish it - should be around December!

[SmedIndy]

Originally posted by Fritz Buelow
Well, WS arrived in the mail today.
727 pages, huh?
Will let you know when I finish it - should be around December!

I got mine today too, but it's behind the Captain Beefheart bio I'm reading, a Pink Floyd bio, and a book on WW I (actually, two books on WW I).

I'm not in a numbers mood right now. I'm stepping out of the triangle, into striped light.

[Fritz Buelow]

Quick sidebar: Smed, try "England's Dreaming" by Jon Savage

[SmedIndy]

Originally posted by Fritz Buelow
Quick sidebar: Smed, try "England's Dreaming" by Jon Savage

Have it in hardcover. Read it when it first came out. Also have Legs' McNeil's book on the New York scene, a book on the LA scene, and bios on a lot of 'em.

Also read Greil Marcus' take on the Dadaists, Situationists, and Punk Rock, "Lipstick Traces", I believe. It's not an easy read, but fulfilling.

OK..enuff...

[gyb13]

:topic: c'mon guys....

Lee, Steve (and others) - gives us updates as you digest through it...

[Baudib]

I got the Win Shares book...


Pretty cool, I think Bill James wanted the book to inspire all sorts of studies and research, and it does look like an invaluable research tool. Frankly it's a little too numbers oriented for me, I prefer the essays, but I recommend it.

It lists Win Shares for all players, all teams....here are some interesting tidbits:

According to Win Shares, the 1941 AL MVP race is, in James' words, too close to call. TW edges Joe D. in WS 42-41. Surprising: 1941 is DiMaggio's best season, according to Win Shares (would have thought it was 1937), but Williams has three seasons better than his .406 year (1942, 46 Win Shares; 1946, 49 Win Shares; 1947, 44 Win Shares. Needless to say, WWII probably broke up the greatest 6-7-year run in history.)

Hank Aaron edges Willie Mays in career Win Shares, 643-642. Mantle has 565, but neither Aaron nor Mays come anywhere near Mantle's high of 51 in 1957, or 49 in 1956, or 48 in 1961.

[Baudib]

Originally posted by nyy26wc


A few more IFs--

IF you believe that a player should be compared to "replacement level" instead of average.

IF you believe that James all of a sudden found the correct way to handle an apparently arbitrary defense system that he tried using in the early 1980s and just discarded.

Do you remember his defensive system from the early '80s, before he lost interest in it? He had a 40 point stat, a 30 point stat, etc., for each position and then assigned points based on the player's defensive stats in each particular stat.

He's revived the same idea, giving no strong argument for how he got right this time.

IF you believe people when they tell you their system works, without them making a strong effort to prove it

I've gotten up to page 100 and don't know when I'll waste my time finishing it.

I guess it all depends on your perspective, but it seems pretty clear to me that he goes to great lengths to explain his methodology, even the runs created formula. Moreover, he gives you the chance to work through the formulas step by step.

In retrospect, I think it is a good thing that the NHBA came out before Win Shares, because James wrote a couple of essays since the NHBA came out, addressing some misconceptions about his system based on reader feedback.

As to your first point, the reasons why comparing players to replacement level instead of comparing them to average are 1. Obvious 2. Numerous 3. Have been debated thoroughly on this board.

Furthermore, Win Shares doesn't use replacement level as a baseline -- this is also explained.

[gyb13]

Neyer likes BJ...ya think? (http://espn.go.com/mlb/columns/neyer_rob/42798.html)
Bill's now come out with two great books in the space of six months, which I presume means we'll have to wait quite some time for the next one (and no, I haven't heard what he's working on next). So this might be a good time to talk about what sets Bill James apart.

For a lot of people, you say "Bill James" and the first thing they think of is "statistics." There's a much smaller group of people who will say, "I know about the numbers, but what's always attracted me to Bill is his writing."

And what I say is, "How can you separate one from the other?" Bill's a brilliant analyst, but there are a fair number of brilliant analysts. Bill's a fantastic writer, but there are a fair number of fantastic writers. What makes Bill special is ... wait, let me back up for a moment here ...

Who was the greatest baseball player ever? Babe Ruth. And why? Because he combined two rare abilities: the ability to win 20 games as a pitcher and the ability to hit 60 home runs as an outfielder.

Who's going to be the greatest baseball player for the rest of this decade? Alex Rodriguez. And why? Because he combines two rare abilities: the ability to play solid defense at shortstop, and the ability to hit 50 home runs.

Who's the greatest baseball writer? Bill James. And why? Because he combines two rare abilities: the ability to write beautifully (in a populist, Stephen King/William Goldman sort of way), and the ability to think about baseball as very few of us have.

Many of you already know all of this. But I present the above as a public service for everybody else reading, because you don't know what you're missing. You really don't.

[Slippery Pedro]

Neyer likes BJ...ya think?

Same as Boo-Boo likes Yogi.

[moose]

Originally posted by gyb13
[Neyer likes BJ...ya think?[/URL]


let's keep this a family site, gustavo.

[KCBOOMER]

I am at a loss to understand some of the invective against James' work here. I for one give him kudos for trying to develop such a system.

Is the system so flawed as to be useless? If so, then harsh criticism is deserved. If it is useful can improvements be made? If so, let's make them and have a better tool.

Hell, maybe it only works for hitters and/or pitchers. I don't know, but I would like to know whether we have something to build on or not.

[richie17]

have a look at fanhome.com 's strategy and sabermetrics forum - Win Shares is generally not well thought of there. I'm not clever enough to follow all of the technical stuff, but it's pretty interesting.

cheers
Rich

[Baudib]

Originally posted by KCBOOMER
I am at a loss to understand some of the invective against James' work here. I for one give him kudos for trying to develop such a system.

Is the system so flawed as to be useless? If so, then harsh criticism is deserved. If it is useful can improvements be made? If so, let's make them and have a better tool.

Hell, maybe it only works for hitters and/or pitchers. I don't know, but I would like to know whether we have something to build on or not.

I think people are annoyed that they have to buy two books to get all of his work. There have been cynical suggestions that it was done in two books in some clever marketing strategy.

I can tell you definitively, that
1. Stats Inc. doesn't really have a marketing strategy. I've dealt with them enough over the years -- they're really sort of a Mom-and-Pop organization -- and there's been a complete overhaul of the company in the past year and a half.
2. The books were done for different publishers. Simon & Schuster did the NHBA, presumably because they published the first.
3. I don't know how you could expect to fit this all in one book. I'm rather glad I don't have to lug one huge book around.

[SmedIndy]

I just read the intro. I'm intrigued, yet still skeptical. I think it may answer some questions and ask more, and even though he says it, I'm not convinced that players on poor teams aren't hurt by win shares.

[Skip]

Originally posted by Baudib
There have been cynical suggestions that it was done in two books in some clever marketing strategy.
I suppose I was at least one of the first cynics. Hey, if the WS book is as big as it it and isnt fluff, I'll be first to admit I am wrong. I still think James could have been clearer in the NHBA without diluting the rest of the text. Could be wrong; I'll wait till I see it to decide.

[SmedIndy]

Originally posted by Skip

I suppose I was at least one of the first cynics. Hey, if the WS book is as big as it it and isnt fluff, I'll be first to admit I am wrong. I still think James could have been clearer in the NHBA without diluting the rest of the text. Could be wrong; I'll wait till I see it to decide.

A lot of the book are the numbers themselves. He could have put the explanation in the NHBA. It wouldn't add THAT much to it.

[Skip]

Originally posted by SmedIndy
He could have put the explanation in the NHBA. It wouldn't add THAT much to it.
If that's true, I go back to my original question about the NHBA (missing features that people loved from the earlier editions) and why he couldn't explain what he was using as justification (at least sort of, minus the subjective make-it-look-like-I-want-it-to factors). Guess I'll have to wait till I get the WS book to form a definitive opinion.

[SmedIndy]

I've read the first few sections of this book. It's hard to wade through, even for a numbers guy. A lot of it does make sense, but I think at times he goes overboard in wanting us to believe his suppositions. Also, he's not a good technical writer, as others have said he's a better essayist.

I do like the defensive system, unfortunately its way too complicated for anyone to figure readily without using all of your Microsoft Office suite and killing your hard drive.

[Rauseo]

yeah it is a good defensive system thanks to Charlie Saeger for comming up with it and publishing it in BBBA 99.

[SmedIndy]

Originally posted by Rauseo
yeah it is a good defensive system thanks to Charlie Saeger for comming up with it and publishing it in BBBA 99.

Is it exactly the same? Or are they just similar? I don't want to accuse James of stealing something, but wondering if they are truly the same.

I had one BBBA and didn't like it that much. I didn't find it as readable as the Abstracts or Baseball Prospectus.

[Rauseo]

Its similar enough that their is no excuse for not citing it, and not knowing about it would hold water in any other field at least as far as I know, though I am no expert.

When reading the Wins Shares book, and Bill went on and on about how excited he was about his system, I was excited too. Until I got to the explanation of the defensive system and instead of the light bulb going on and me understanding I was thinking, how is it that james just know is understanding this? I got it 3 years ago. Part of the article is on the bbba website:

http://www.bigbadbaseball.com/homepage/defense.html

judge for yourself. The key is to paraphrase Bill looking at defnese from a top down approach. Which clearly is what Charlie is doing.

[wyvern37]

Originally posted by Rauseo
yeah it is a good defensive system thanks to Charlie Saeger for comming up with it and publishing it in BBBA 99.

Ummm...he mentions that he got several key elements of his idea from Bill James work. How can James be stealing from him when Charlie got his ideas from James in the first place?

Karim

[pwdennis]

WIN SHARES has rather less text than is typical for a Bill James production. For this reason, fewer readers will want to read it and the major book chains (Borders, Books-A-Million, Barnes & Noble) are not stocking the book, although all will special order it for you.

I am about 75% through the book and I do like it. Although qualitative analysis of defensive stats, particularly for infielders, has shed light on the defensive value of the various players, there are still difficulties in assessing outfielders as not enough notice has been given to the effects of the different ballparks on putouts. For example, a pair of prominent left fielders of the 60s/70s , Pete Rose and Carl Yastrzemski are rated B- and C+ respectively. While Rose may indeed be a B- as a fielder, there is almost no one who would regard Charlie Hustle as a defensively superior LF to Yaz. Fenway, with its limited area and Green Monster, tended to limit putouts. Contemporary observers regarded Yaz as a superior fielder and he won a number of GGs. Similarly, I do not buy Winfield as a D+ outfielder.

It should be noted that the analysis of defensive data is a constantly evolving process. While James does not have it completely right yet, at least many of the absurdities of linear weights have been avoided.

My suggestion for Bill James in compiling ratings would be to determine WIN SHARES per 162 games and determine the average.

I look forward to his next book

[maildave]

Haven't read it, and beyond this question, I'm not even very curious about it.

Bill states in the other new book - the one I actually own - that Win Shares will always total the number of wins a team has over a season times three. Period, end of statement. If a team wins 100 games, they'll total 300 Win Shares.

Is this based on the pythagorean record, or the REAL record? If it's the real record, I just read something by Mr. James on Tom Tippett's Diamond-Mind site saying that one-run wins are just about all luck.

That being the case, wouldn't Win Shares be skewed an increasing amount for each one-run game a team plays?

If it's based on pythagorean, then nevermind. :)

DJ

[sweaver]

It's based on the actual record. This is one reason the book is so controversial.

[maildave]

I'd have brought that up during the super-long Bill James chat on ESPN a week or so ago, but I'm sure he would have been sarcastic towards me somehow.

But hey, I love the other book - I just have to discount the Win Share stuff in it.

DJ

[Skip]

Just pushing this up - because it is worthy of debate.

Is Win Shares legit?
With or without the subjective part?
Is it better than prior measures of offensive value?
Of defensive value?
Of pitching value?
What changes might improve it?

(Lets talk about something meaty; I'm tired of Canseco! :D )

[wyvern37]

I think the question might be to what degree is it legit.

I think that Win Shares is an excellent stat for three reasons:
1) It is based on established statistical measures like Runs Created and Constituent ERA.
2) It normalizes them in the context of the team and stadiums.
3) It includes defensive value in the rankings.

The last item always gets the most criticism. I cannot for the life of me understand why. First of all, it is far more accurate than Total Baseball's ratings and a whole lot better than nothing at all. And secondly, most of their defensive Win Shares are a result of the position they play anyway.

[calexpat]

I haven't read the book, but a simple question about the idea has been rattling around in my head. If a Win Share is literally about assigning a team's actual wins among its players, shouldn't all stats that occurred in losses be ignored? Does James do that?

[SmedIndy]

No. He's trying to break down the components of a players contribution to the team, and dividing them on how well the team does.

As far as the defensive part, I think they're highly accurate, yet maddeningly complex to calculate. Part of the beauty of baseball stats is that you could do most of them with pen and paper.

[poorme]

Originally posted by Skip
Just pushing this up - because it is worthy of debate.

Is Win Shares legit?


I think Win Shares is the best available measure. I wish he had left out that subjective aspect, though. The defensive measure makes a lot of sense, although it's kind of hard for Jo Shmo to understand. Not that Jo couldn't figure it out, he just doesn't have the inclination.

[calexpat]

Originally posted by SmedIndy
No. He's trying to break down the components of a players contribution to the team, and dividing them on how well the team does.


Does that mean you can only calculate Win Shares for large numbers of games?

I go three for five with a two-run homer, a stolen base, and a fine running catch on opening day, but my team loses 6-5 in the ninth. How many Win Shares do I have?

I go 1-5 in my first 6 starts, with 5 quality starts and an Adjusted ERA of 2.86. My one win is a 4-3 affair, with player X hitting a pinch hit grand slam in his only at-bat of the season. Meanwhile, the rest of the staff is 12-8 with an adjusted ERA of 3.82. The team, of course, is 13-13. How many Win Shares do I have? How many Win Shares does player X have? How many Win Shares does the rest of the staff have?

It seems to me that if a Win Share means, literally, contributions to wins, you have to completely ignore performance in games lost.

[SmedIndy]

No.

Win shares is staggeringly complex, especially the fielding part.

However, he determined that each team has X amount of win shares, and they are 3 times the number of wins a team gets.

This works, seemingly, for every team in baseball history, except for some of the early teams that were historically egregious.

Because the data does not truly exist on a game by game basis to the depth it is required, it would be foolish to count just the "won games".

A player can help his team in attempting to win (which is all a player can do, really) and they lose due to things that are in no way related to his performance.

[poorme]

Originally posted by calexpat

It seems to me that if a Win Share means, literally, contributions to wins, you have to completely ignore performance in games lost.

You could easily adjust the formula to make it "loss shares", if that makes you happy.

Your method might be an interesting curiosity... it would penalize players who played on bad teams. Theoretically if Barry Bonds' team played a whole year witout winning a game, he would get 0 win shares. Doesn't measure his ability, though.

[calexpat]

Originally posted by SmedIndy
A player can help his team in attempting to win (which is all a player can do, really) and they lose due to things that are in no way related to his performance.

Of course, I think performances in losing efforts have value. But it seems to me that the very idea of Win Shares implies that they don't. Perhaps the name is misleading?

Anyway, I didn't read the book, so I'm not going to dig in against it. And I'm certainly not proposing an alternative method. I was more going for a reductio ad absurdum of the concept behind Win Shares.

[VNV Nation]

You can't calculate Win Shares for a single game. A team gets 3 Win Shares per team, and then they're divided up among the players. The distinction between 1 and 2 Win Shares per player is hardly significant, and getting into fractions of Win Shares is downright absurd.

[VNV Nation]

Originally posted by calexpat


Of course, I think performances in losing efforts have value. But it seems to me that the very idea of Win Shares implies that they don't. Perhaps the name is misleading?

Anyway, I didn't read the book, so I'm not going to dig in against it. And I'm certainly not proposing an alternative method. I was more going for a reductio ad absurdum of the concept behind Win Shares.

Win Shares are essentially Win Created. He does not get into specifics of "this sac fly contributed to .20 of a Win on 6/20" but, rather, extrapolates things we do know already (relationship between runs and runs allowed is similar to Winning Percentage, etc.).

[calexpat]

If that's the case, isn't it somewhat arbitrary and misleading to use actual wins, rather than expected wins?

[VNV Nation]

No, the idea is that whatever happens on a ballfield, someone gets credit for it. If it's based on expected wins then essentially you're pulling wins out of the air.

[calexpat]

But if the result refers to actual wins, then the method ought to as well, right? As it is, the wins may not be "pulled out of air" of air, but the method of apportioning them is, it seems to me. Not that there's anything wrong with pulling things out of air. I just think that by using actual team wins as the basis for Win Shares, James implies that individual Win Shares refer to actual wins, and they don't.

In practice, I'm sure this is a very small problem. But I think it's a real conceptual problem. Not necessarily a soluble one, though.

[VNV Nation]

I'm not sure what you mean. Win Shares does refer to actual wins. If the Giants win 90 games, their players have 270 Win Shares, if Barry Bonds has 48 Win Shares, that means he won 16 games for the Giants.

A radical departure from other methods, Win Shares starts with what the team accomplished and then seeks to proportion credit (through a myriad of adjustments) correctly.

[calexpat]

I can see the Giants' 90 wins, but I can't see Bonds' 16 wins. If the problem with sophisticated statistical analysis is that it seems to pull runs, wins, etc. out of thin air, I don't see how Win Shares avoids that problem in any significant way.

I'm sure it's an effective system. It just seems misleading in the sense that it claims to be more than the sum of its parts (ie, the various methods of analyzing offense, pitching, and defense, plus the Pythagorean method for relating runs to wins).

This is starting to go around in circles, and I do feel uncomfortable opining at such length without reading the book, so I think this will be my last post on the matter.

[poorme]

It's called Win Shares because it takes actual wins and then "assigns" them to each player on the team based on performance. Three points.

A. You couldn't get historical stats for wins only.
B. What does it matter since it should not have any effect on the player's ordinal ranking? This is the key -- given a decent sample size player ranking should not change if you look at performance in all games or just wins.
C. A pitcher's performance in one game has an effect on other games, regardless of the win loss situation.

[Sopalt]

I just got the book yesterday. Although I haven't gotten very far into the book, it certainly does appear interesting in terms of this new system of evaluating players past and present. However, I must say that I am most unimpressed by this system.

Although I am the type of person that hates when people make arguments by "gut feeling" without any real support to back it up, I will do such a thing. Does anyone believe that Frank Tanana, Jerry Koosman and Rick Reuschel were superior pitchers to Randy Johnson? Yet, that is precisely what James' system asserts. Through last year, Johnson had 226.4 Win Shares. Reuschel had 239.8, Koosman 241.0 and Tanana 241.4.

Let me use Reuschel as an example. He pitched from 1972-1991, had a record of 214-191 and an adjusted ERA of 114. Johnson, before this season, pitched from 1988. His record was 200-101 with an adjusted ERA of 140. He also won four Cy Young awards, finished 2nd twice and 3rd once. Reuschel never won the Cy Young award nor even finished second in the voting. He finished 3rd twice. Also, using other tests that James developed and wrote about in one of his previous books, Johnson has 69 Black Ink points to Reuschel's 7. For Hall of Fame standards, Johnson has 52.0 points to Reuschel's 31.0. How then, can anyone say that Reuschel had a better career than Johnson had through 2001?

So as to show that I haven't just picked one "anomaly" to pick at, how about these others? How many people would say that Bret Saberhagen had a better career than Sandy Koufax? How many would say that Jerry Reuss was superior to Koufax? How many would say Chuck Finley has had a better career than Koufax? How many would say that Tommy John and Jim Kaat were superior to Whitey Ford? How many would say that Dennis Martinez was superior to Randy Johnson? How many would say that Bert Blyleven was superior to Jim Palmer?

Basically, this leads me to question the credibility of James' system. If two players are comparable, I can see some choosing one as better than the other and other people the reverse. However, to me, there is no way that Reuschel is even comparable to the Big Unit, much less that he should be rated #67 among starting pitchers and Johnson #84. Heck, Pedro isn't even ranked among the top 150 pitchers to have ever played the game!

My dad was an "old school" engineer. At the end of his career, he wasn't computer-literate. Yet, from experience, he could tell if a younger engineer's calculations were in the ballpark or not even in the right county. His favorite saying, when it came to computers was "Garbage in, garbage out", meaning that the young engineers, with all their fancy computers and equations, if they didn't know what they were doing, were doomed to come up with bad results. My suggestion is that for all the fancy mathematics that James expounds in his latest book, he should do "real world" checks to see if he's getting decent results or not.

[Skip]

In the Bonds = Williams thread in Baseball History, PWDennis and I had the following exchange (italics added by me):

Originally posted by pwdennis
The drumbeat for Barry Bonds has been building for a number of years. In his write up for Bonds in THE NEW BILL JAMES HISTORICAL ABSTRACT, James does not have Bonds incredible 2001 season in the mix yet says of Bonds (and had been saying for several years prior) " Certainly the most un-appreciated superstar of my lifetime... Probably the second or third best hitter among the 100 listed left fielders (behind Williams and perhaps Musial), probably the third best baserunner (behind Henderson and Raines), probably the best defensive left fielder. Griffey has always been more popular, but Bonds has been a far,far greater player...."

Originally posted by Skip
And then James in Win Shares during his Random Essay Win Shares Defensive Rankings Expressed as Letter Grades gives Barry a 'B-'. I understand that LF without enough overall skill to make his list of the top 100 may still deserve to be called better fielders, but to call him the best defensive LF of the overall top 100 LF, and then grade him 'B-' on defense is just maddeningly inconsistent.

Originally posted by pwdennis
But James himself points out that using strict statistical grading creates some problems that still need to be sorted out, particularly when it comes to outfielders. For instance Win Shares has Pete Rose (B-) rated ahead of Yastrzemski (C+) defensively, largely because Fenway greatly reduces LF putouts. No one who saw Rose and Yaz would ever pick Rose ahead of Yaz as a defensive player, nor would they pick Rose ahead of Dave Winfield whom Win Shares calls a D+ outfield.

I don't hate James, or even dislike him nearly as much as this next statement will make it sound ... but I wish I could write two books based on a stupefyingly detailed objective numerical means of analysis, which is then adjusted by a barely explained subjective component, and then identify that the system doesnt really work in the same books, and still get published and make money.

[VNV Nation]

Sopalt:

I assume that you hadn't read the New HBA. He could have explained it better in the actual Win Shares book, but doesn't. Bill writes that Career Win Shares is not some magic number rating, like TPI in Total Baseball. Win Shares is like wins, or runs, and he points out that Mark Langston is a better pitcher than Sandy Koufax, just because he had more wins. Reuschel pitched a lot longer than Johnson (coming into this season), and Johnson had a lot of ordinary seasons at the beginning of his career.

In the HBA, career Win Shares is just a fraction of the formula used to rank players, along with top 3 Win Shares seasons, best five consecutive years, and Win Shares seasonal notation. Johnson is ranked (not counting 2001 season) as the 49th-best pitcher of all time, with the caveat that all active players (except Biggio!) are ranked as low as they possibly can be, while Reuschel is 81. Pedro Martinez is 29th and Tanana and Koosman don't make the top 100.

[VNV Nation]

Skip:
re: Bonds' grade, all outfielders are clumped together, so all of the "A" grades go to the good center fielders:

"A rating can be outstanding for a left fielder, but unimpressive compared to a center fielder. But it's what we can do with what we have."

I don't believe that there's a true corner outfielder that gets better than a B-. It would have been nice to sort the outfielders by position, and then grade them accordingly, but he doesn't do that. Too much clutter in trying to find out where each guy ranks, I suppose...breakdowns per position weren't available when he started the book (TB came out with that a couple of years ago).

[Skip]

Originally posted by VNV Nation
Skip:
re: Bonds' grade, all outfielders are clumped together, so all of the "A" grades go to the good center fielders:Good point. It still rankles, but ... ok.

[VNV Nation]

Don't be rankled, play chess.

[pwdennis]

Originally posted by VNV Nation
Skip:
re: Bonds' grade, all outfielders are clumped together, so all of the "A" grades go to the good center fielders:

"A rating can be outstanding for a left fielder, but unimpressive compared to a center fielder. But it's what we can do with what we have."

I don't believe that there's a true corner outfielder that gets better than a B-. It would have been nice to sort the outfielders by position, and then grade them accordingly, but he doesn't do that. Too much clutter in trying to find out where each guy ranks, I suppose...breakdowns per position weren't available when he started the book (TB came out with that a couple of years ago).


Quite true, however the fact is that CF is the most consistant of the outfield positions in terms of symmetry. Outside of the Polo Grounds with its endless deep center , most of the really eccentric field dimensions were in the corners - think of Forbes with its odd angles, Fenway with its Green Monster, Polo Ground with its short foul lines, Ebbets Field, Oakland with its endless foul territory or countless other oddities and relics. It is in the corner outfield positions that James methodology breaks down. Undoubtedly, in future editions, the method will be refined to make sense out of the current chaos.

[Sopalt]

Upon further review of the book, I'm no longer unimpressed by Bill James' new system, but rather hate it. In the Introduction, James says that Win Shares is basically Wins Created. Obviously James has never played baseball or for that matter any team sport. Whatever happened to the saying, "You win as a TEAM and lose as a TEAM"? The fact that James tries to say that Player A was responsible for his team winning x number of games while Player B was responsible for his team winning y number of games is simply absurd. For this reason, I sure hope that his system doesn't catch on in popularity.

Isn't it enough that as fans, we have to watch a number of today's players playing only to pad their statistics which lead to bigger contracts? If it is accepted that it is possible to determine exactly how many games a player won for his team during the previous season or two, this will lead to players using this as bargaining chips to get bigger contracts. Soon, guys won't be concerned as much about the team, but rather how many wins he gets credit for.

There's another thing that is overlooked in James' system. A friend of mine once said "Baseball is a game of failure. What matters is what someone does with that failure." That's so true. Ted Williams was the last guy to hit .400 in a season. Even during 1941, that meant that he failed to get a hit approximately 59.4% of his at bats. Who are the "unsung heroes" that managers love? Aren't they the guys who do things to help the team win, things that don't necessarily show up in the box score? How does a system account for a guy advancing the base runner by hitting behind him? Let's say the score is tied and a guy sits on second with no one out. Yes, the ideal would be for the batter to get a hit to bring the guy in. Chances are, however, that the batter will not do that. The batter tries to pull the ball and hits a grounder to the third baseman who throws him out at first. One out and the runner is still on second. It still takes a hit to bring him home. However, what if the batter had hit a ground ball to the second baseman? He still gets thrown out at first, but the runner moves to third with one out. Now, in addition to the next batter getting a hit, he can score on a fly ball, a passed ball, a wild pitch, a squeeze bunt, or even a ground ball out, provided the other team isn't playing their infield in.

This is just one of a number of things that affect the outcome of a game that can't be accounted for by any system. That's what makes baseball a beautiful game. It's true essence cannot be captured by mathematical formulas.

[sweaver]

All those things can be captured mathematically, and that's the beauty of baseball. Bases advanced can be accounted for. How do you quantify a block in football, or a pick in basketball? In baseball, all these things are part of the record. It is just a matter of how sophisticated you want to make the numbers.

Win Shares is not perfect. But let's understand the purpose: to sum up, as best as possible, the contributions of each player into one number. A daunting task, to say the least.

[gyb13]

Originally posted by Sopalt
If it is accepted that it is possible to determine exactly how many games a player won for his team during the previous season or two, this will lead to players using this as bargaining chips to get bigger contracts. Soon, guys won't be concerned as much about the team, but rather how many wins he gets credit for.
the reality is the same right now, only instead of 'wins', it's 'home runs' or 'saves'. you can't eliminate individual accolades from contractual negotiations unless you eliminate individual statistics all together. wouldn't baseball be better if roberto hernandez wasn't making 6 mill a year based on a meaningless statistic?

[sweaver]

Finally got my copy today. A quick perusing shows many interesting factoids and arguments. I like the system overall, but it certainly has flaws. The most glaring I have seen: Richie Zisk is rated a C+ in the OF for defense. So is Carl Yastrzemski. I think there's something wrong with this picture.

[sweaver]

Now that I have read the whole book, my observations on the latest tome from Bill James:

This is a hefty volume at 700 pages plus, even in paperback. The first 260 pages include explanations of the system and a look at some of the applications, while the last 2/3 is listings of Win Shares in tables for reference purposes. There are writings on how credit is earned, how Win Shares points (3 for each team win) are divided between hitting, pitching and defense, and how Shares are "claimed" by individual players. The hitting is based mainly on the seasoned Runs Created formulas, with a correction for the actual runs scored as to calculated runs created. Pitching is based mainly on runs saved and innings pitched, with extra credit for relief innings (usually higher-leverage) as well as for wins and saves.

The fielding is a complicated system, since even the best defense can only record 27 outs in a 9-inning game. Therefore, certain assumptions are made that good teams have a good defense, and decisions to assign points are made accordingly.

James then applies the system to various arguments, such as who should have won Gold Gloves and MVP awards, was this guy better than that guy, and such.

My view? This is an important milestone, but by no means a be-all and end-all. James has been trying to encapsulate a player's performance into one number for years, starting with "Approximate Value" in the old Abstracts. That was, of course, approximate. This is an attempt to assign an EXACT value to a player for each season. It is a noble effort, but falls short in many ways, and will doubtless undergo much revision in the years to come.

Particularly, the defensive rankings will change over time. I think James has a good framework, but he has fallen back on his 40-30-20-10 point system he has used before, rather than evaluate on a more fluid level. Much of the work he seeks to do here has already been done by Baseball Prospectus, and done better.

I like it. I like it a lot. But there is a ways to go yet.

[Craig S.]

Originally posted by sweaver
I like it. I like it a lot. But there is a ways to go yet.

I agree that there's a ways to go, but it's a fantastic book. The figures he comes up with are fodder for countless discussions, and I found it tough to stop reading the "Random Essays" part, which I found fascinating.

[Jim Rice]

I've read about a third of Win Shares, so I've not done a complete study of it yet and I'm not too much of a numbers guy anyway, but I do have a couple of initial impressions.

First, I think his defensive numbers are soundly based, particularly compared to Total Baseball. I can't comment on whether or not he borrowed some of the work from others, but if he did, it looks like he at least borrowed the right stuff. True, the corner outfielder rankings need work (and am I mis-reading it or did he give Bonds a C, not a B- as some have stated here?), but otherwise the defensive numbers make sense.

I haven't had a chance to dig into the pitching details too much yet, but so far they seem solid enough.

The methodology on the hitters makes sense with two fairly glaring exceptions:

1. His Runs Created formula factors in raw GIDP totals in lieu of GIDP rates. GIDPs, like runs and RBI, are a TEAM function, not an individual one. You simply cannot hit into a double play without a teammate getting on base in front of you and a manager not ordering a steal or hit-and-run to avoid the double play. Yet the hitter gets full "blame" for the GIDP. All other such "team" numbers were excluded from the Runs Created calculation - why include this one? If you could devise a GIDP rate (GIDP/GIDP Opportunities) and include it, I wouldn't have a beef, but James doesn't do that. Maybe the data simply isn't there, but in that case he should have probably excluded them entirely.

2. He adjusts for era and he adjusts for ballpark, and in the defensive calculations he accounts for the left- or right-handedness of the pitching staff, but James fails to account for the left- or right-handedness of the HITTERS. Right-handed hitters are at a platoon disadvantage at least two-thirds of the time; for lefties the reverse is true, they're usually facing a platoon advantage. Is this the hitter's fault? Of course not. You can't MAKE a player right-handed or left-handed. You can teach a youngster to hit from the left side, but that's a function of good coaching. Should a player really be penalized for not having good coaching in their youth? Obviously not, which means there should be some adjustment made for righties in this system before they can be evaluated evenly against lefties (to say nothing of switch-hitters). That seems to be a pretty gaping hole that isn't even addressed in the parts I've read so far. Maybe he does so later and I just haven't got that far yet. I certainly hope so, otherwise he's essentially under-valued every right-handed hitter in history.

[poorme]

Originally posted by Jim Rice

2. He adjusts for era and he adjusts for ballpark, and in the defensive calculations he accounts for the left- or right-handedness of the pitching staff, but James fails to accounts for the left- or right-handedness of the HITTERS.

what's the difference which way they bat? it's the stats that matter. i don't get it.

[Skip]

Originally posted by Jim Rice
am I mis-reading it or did he give Bonds a C, bot a B- as some have stated here?As the 'some' being referenced here, I humbly say oops. I must have read across the Bobby Bonds line (B-) instead of the Barry line (C). Of course, that makes my original complaint even stronger, despite the lumping of all OF so only CF get A's.

[Jim Rice]

Originally posted by poorme


what's the difference which way they bat? it's the stats that matter. i don't get it.

The difference is enormous, and is the entire basis for platooning players. Righties hit lefties better and vice versa. That's essentially been a given for about fifty years. What it means from a statistical standpoint is that all righty hitters compiled their career numbers while at a platoon disadvantage for at least two thirds of their plate appearances. They cannot be compared to a lefty or to a switch-hitter without normalizing for this difference.

It's the same as ballparks - it's much more difficult to hit 300 homers in Chavez Ravine than it is in Coors Field. We normalize for that. It's the same for era - it's much more difficult to hit .301 in 1968 than it is to hit .340 in 1930, so we normalize for that too. Well, by the same token, it's much more difficult to post a .900 OPS when you're at a platoon disadvantage 70% of the time than it would be if the ratio was reversed, so why in the world don't we normalize for it?

This is of particular signifigance with Bill James because he recognizes all of the other normalizations that need to be made and, in fact, recognizes that left- or right-handedness matters in pitching and defense, but he ignores the platoon advantage entirely in evaluating hitters. It's an inconsistency that he decided not to explain (at least, I haven't come across it in Win Shares yet).

[poorme]

I think what we have here is a failure to communicate.

I thoroughly understand platoon advantages. What I don't understand is your point. let's compare ichiro and sweeney. they're both hitting around .340, one's a righty and one's a lefty. let's say their stats happen to be identical. why should one get more win shares than the other?

[Jim Rice]

Unfortunately, you've picked two guys who are each in the minority and actually hit righties and lefties equally well. As you know, that's rare.

To illustrate my point, I'll pick two different guys. Magglio Ordonez, a right-handed hitter, has an overall OPS of 989 this, but that's made up of an OPS of 1031 against lefties and a 979 OPS versus righties. Unfortunately for him, he only gets to face a lefty in 21% of his at-bats. If that ratio were reversed, as it is for lefty hitters, and he got to face a platoon advantage almost 80% of the time, his overall OPS wouldn't be 989 - it would be 1021.

Now let's do the reverse on a lefty hitter - Jason Giambi has an OPS of 1014, but that's made up of an OPS of 1063 versus righties and just 893 versus lefties. Fortunately for him, he only faces lefties in 29% of his at-bats. If that ratio were reversed and his performance remained unchanged, his overall OPS wouldn't be 1014, it would be 920.

In short, if Giambi had to face a platoon disadvantage that approximated the frequency Ordonez faces one, it's likely he'd post an OPS significantly lower than Ordonez's - 920 versus the 989 Mags is putting up. As is, his OPS is higher than Mags' and he's considered a better hitter as a result. Now, obviously this is a one-year example only without park factors and other factors accounted for - I'm NOT arguing that Magglio Ordonez is a better hitter than Jason Giambi, overall. But I am saying the Magglio Ordonez's number don't give a true reflection of his ability because he faces a disadvantage much more frequently that lefty hitters do.

And that's the point - James has ignored this aspect entirely in Win Shares. He makes the claim that a defender can't be fairly judged unless we consider who is doing the pitching - a righty or a lefty. BUT HE IGNORES THE PERSON HITTING AGAINST THAT PITCHER. If the arm used to throw the ball affects how it is hit - thus affecting who fields it - why on Earth wouldn't that be reflected in the evaluations of the people who actually do the hitting?

[poorme]

Originally posted by Jim Rice
But I am saying the Magglio Ordonez's number don't give a true reflection of his ability because he faces a disadvantage much more frequently that lefty hitters do.


Ooooh. Now I see what you're saying. That's what makes sweeney's performance (and MVP era frank thomas) all the more amazing. right handed hitters without much speed.

anyway, i don't know of a way to solve that problem. has anyone computed platoon splits pre 1985?

[Jim Rice]

Originally posted by poorme

anyway, i don't know of a way to solve that problem. has anyone computed platoon splits pre 1985?

Retrosheet's got them from 1974 through 1990 on their site, plus AL only for 1967-1969. Otherwise, I don't know who would have a more comprehensive data source - if anyone does, it's probably James and the STATS people.

Like I said, if the data simply isn't there to support the calculation, so be it. It just would have been nice if he'd come out and said so.

[poorme]

It would be nice if james or ESPN or someBODY would provide the win shares for last season. james is dropping the ball here....(unless they are under my nose somewhere).

[VNV Nation]

Rice, I don't know what you're saying here. Are you saying that the handedness of the batters faced should be factored into defensive evaluations? If so, then I agree, sorta. also, left-/right-handedness of the pitching staff should be taken into account when evaluating left and right fielders, but they're all clumped together.

much of this information isn't available though. but the handedness of the batters is already reflected in the system, because the handedness of the pitchers is in there. get what I'm saying? if a lefty gives up 13 grounders to short, and a righty only 12, that's due, presumably, to seeing more righty batters, and it's accounted for.

as far as giving righty batters extra credit for dealing with the platoon differential more often, there's no point to that. the only thing that matters is results. that's like saying rebounding in the NBA is influenced heavily by size and strength, and thus, Muggsy Bogues is a better rebounder than Shaq, because he's not nearly as big.

to be clear, Win Shares NEVER measures ability, it measures value.

[sweaver]

Originally posted by poorme
It would be nice if james or ESPN or someBODY would provide the win shares for last season. james is dropping the ball here....(unless they are under my nose somewhere). Well, you notice I've been putting up the ones I have, which is just the leader boards really. STATS Inc is figuring them, so they must be available, I just don't know where they will be published.

[Jim Rice]

Originally posted by VNV Nation
as far as giving righty batters extra credit for dealing with the platoon differential more often, there's no point to that. the only thing that matters is results. that's like saying rebounding in the NBA is influenced heavily by size and strength, and thus, Muggsy Bogues is a better rebounder than Shaq, because he's not nearly as big.

to be clear, Win Shares NEVER measures ability, it measures value.

But it is inherently unfair to right-handed hitters to judge their value versus lefty hitters or switch-hitters unless the platoon advantage/disadvantage is taken into account.

Here's an example - two outfielders, playing for the same team, post IDENTICAL offensive statistics. They should each receive an equal number of offensive Win Shares, correct? So they're equally valuable offensively according to Win Shares. But what if one of those batters produced these numbers facing a platoon advantage in 80% of his plate appearances - typical for a lefty - and the other produced those numbers facing an advantage just 35% of the time - typical for a right-handed hitter? Whose accomplishment is more impressive? The righty. Who is harder to replace? The righty - it's simply harder to find someone who can produce those numbers facing that high a percentage of disadvantage.

The righty is the better hitter, period. In fact, a righty hitter with 17 or 18 Win Shares is probably just as good a hitter as a lefty teammate with 20 or 21 Win Shares because he did so under much more adverse conditions. Win Shares may be designed to measure value and not ability, but that's not how it's being used in many cases. Instead people write "Player X had more Win Shares than Player Y over this span, so he was a better player." That's not always correct. More valuable? Maybe. But "better"? Not necessarily.

[sweaver]

That's an interesting theory, but since the situation you describe is endemic to the game, and cannot be changed (short of a rule requiring equal numbers of left-handed and right-handed players on each team, or something) I'm not sure it has any validity. The situation is what it is.

[Jim Rice]

Originally posted by sweaver
That's an interesting theory, but since the situation you describe is endemic to the game, and cannot be changed (short of a rule requiring equal numbers of left-handed and right-handed players on each team, or something) I'm not sure it has any validity. The situation is what it is.

True, and variations in ballpark dimensions are endemic to the game, but we can measure those affects so we adjust for that to be fair. The number of lefty pitchers on a staff are endemic too, but that affects the defensive numbers of the team's defenders and we can measure it so we adjust for that too. The number of games in a season is endemic, but it's measurable so we adjust the numbers for older players who played shorter seasons. In the sixties, the high mound was endemic to the game and pitchers dominated, but that could also be measured and we adjust for that now also. All of this is done to level the playing field and allow us to measure and compare the value of players of different types, eras, and locations.

The percentage of times a righty or lefty hitter faces a platoon advantage can be measured. Their performances in those situations can also be measured. Why WOULDN'T we adjust for it to level the playing field even further?

[VNV Nation]

Sorry Jim Rice, but this line of reasoning should just be rejected out of hand.

The reason we would adjust for ballparks is because ballpark effects are illusion. Ballparks alter the run-context of the game. A hitter in Coors Field who bats .350 may have no bigger impact than a .290 hitter in Safeco. But a .350 hitter who's right-handed has exactly the same impact as if he were left-handed. A home run from Barry Bonds is just as valuable as a home run from Manny Ramirez.

Following your logic, this would lead to nowhere. If you compare Mickey Mantle to say, Jimmie Foxx, and conclude that Mickey Mantle had it much easier because he never had a platoon disadvantage, would this lead you to prefer to have Jimmie Foxx over Mantle? In the real world, no, of course not, and in fact, you'd prefer to have Mantle, because he's a switch-hitter and because of the advantage that gives you in lineup construction.

It's simply a fact of life that left-handed players have an advantage in baseball, just like tall players have an advantage in the NBA. If you're a left-handed batter and can't hit lefty pitchers at all, you can still be a minor star, like Andy Van Slyke and Jim Thome in his early years. If you're a right-handed batter who can't hit righties, well you're going to be Chris James. James's disadvantage does not make him as good as Andy Van Slyke; in fact, the opposite is true: Van Slyke's left-handedness makes him a better player than Chris James.

[Jim Rice]

That's not what I'm arguing. Andy Van Slyke with NO hands would be better than Chris James. They are unequally talented.

What I'm arguing is that between players of opposite-handedness but similar production, they may be of equal VALUE, but they are not of equal TALENT. As you admit yourself, it's harder for a righty to hit well in the bigs. If he posts numbers that are similar to a lefty hitter, the righty is probably a BETTER player, meaing a more gifted one. He accomplished a similar feat under much more diffcult circumstances.

I understand the previously made point that Win Shares is designed to measure value not talent, and that's fair. But what's not fair is how people apply Win Shares to assess talent anyway. And among those who do it is Bill James. The overwhleming majority of the formula he used to rank players in The New Abstract was based upon the number of Win Shares they accumulated - in total, in a five-year span, their three highest figures in any given year, their Win Shares per game, and so on. Well, if the tool wasn't supposed to be used to measure talent, why does he then do exactly that?

To use your basketball analogy, who is the better rebounder, a 7'2" guy who averages 12 boards or a 6'7" guy who averages 11? (In other words, Wilt or Rodman?) Easy, the shorter guy is a better rebounder. Well, in baseball, who accomplished the more difficulty feat, the lefty who compiled a five year run of 26-26-32-30-27 Win Shares or the righty who compiled 24-24-30-28-25? Probably the righty, but James' system will rank the lefty higher. More productive? Yes. But BETTER? No.

[SmedIndy]

JR - He IS measuring value. There is a disconnect between talent and value in any professional sport. Talent is the ability - value is how you use it.

The lefty put more runs on the board - he added more value to his team - he deserves to be rated higher. Life's like that sometimes.

[Jim Rice]

Originally posted by SmedIndy
There is a disconnect between talent and value in any professional sport. Talent is the ability - value is how you use it.


And all Im saying is that the righty is USING more talent, but the circumstances he is facing are much more difficult so the result often appears that he is no more talented than a lefty who posts equal numbers.

I've beaten this horse enough to I'll leave it at this - If a lefty and a righty both post 40 Win Shares one year and are the only candidates for MVP, I'm voting for the righty for every single time. His 40 Win Shares represent a greater accomplishment because of the conditions he faced to do it.

[SmedIndy]

Parks can affect lefties and righties differently too, so the righty may NOT NECESSARILY be in a "much more" difficult spot. In fact, in some circumstances, it could favor a certain type of righty hitter over any lefty or other types of righties. We could assign values for fly-ball righties hitting in Fenway, versus ground ball lefties hitting in the BOB.

This is all a fine tapestry woven together, and I think us stat guys sometimes think TOO much on the minutae.

[VNV Nation]

just to add my final :2cents: to this issue...(i'd hate for this thread to get closed, because there's lots of Win Shares stuff we could be talking about all winter)....

i don't think you can really measure anything other than value, unless you're analyzing talent in trying to acquire players, in which case it would be legitimate to ascertain "what is this guy's ability, because his value for that team might be considerably different than his value to us."

but really, talent, unless it leads to different results, is useless in evaluating a player. Bo Jackson had more talent than David Eckstein. David Eckstein is a better baseball player.

and the guy getting 12 rebounds is better than the guy getting 11 rebounds, no matter how tall he is.

[Jim Rice]

What I think this all boils down to is that we need to start thinking about evaluating types of hitters independent of other types of hitters. We do that with pitching and defense already, but there hasn't been much effort directed in determining the value of a leadoff hitter versus a slugger, or a lefty versus a righty, etc. Your point about Bo Jackson and David Eckstein is a perfect illustration. Forget the talent disparity, these aren't two players that can be fairly compared to one another anyway. They have wildly different skills sets.

To get back to the hoops analogy, the 7'2" center with 12 boards and the 6'7" small forward with 11, which team do you think is going to do better at rebounding? My guess is that the small forward's team will be a better because it's a lot easier to find an average center who will grab about 9 boards, giving them a total of 20 from those two spots, than it is to find a small forward who comes anywhere near 11 per night. It's more likely they'll find one who grabs about 6, so they'll only be getting a total of 18 or so. In that context, the smaller player with the slightly lower total of rebounds can be argued as having more value to his team. His skill set is much more rare, his accomplishments are much more difficult. Any team with him on the floor has a distinct advantage other teams won't have.

Baseball has similar differences in value that are not easy to quantify in a formula. The ability to be a consistently productive right-handed hitter is more rare than the same ability from the left side. (And to the point about Foxx vs. Mantle earlier - I would argue that Mantle's skills set is so much more rare than he is obviously more valuable to his team than Foxx.) There is value in that that no formula to date even attempts to quantify. I think someone should attempt to do so. That's all I'm trying to say.

On that note I'm done, I swear.

[Skip]

Originally posted by Jim Rice
What I think this all boils down to is that we need to start thinking about evaluating types of hitters independent of other types of hitters. Just how in heck do you assign 'types'? As an extreme example, just what type of hitter is Barry Bonds. BB of 1991 and BB of 2001 are from different planets. I get the idea of where you're coming from, but to do anything like this or any other talent v. value thing just makes the analysis too unwieldy. Heck, it's taken decades for something as simple as OBP and OPS to be broadly accepted.

But ... it's an ivory tower enough idea that you could probably guarantee tenure with the volume of pubs you could produce on the subject. :p

[sweaver]

I think the difference is, there are a certain number of RH and LH people. Now, there are outliers, like Brooks Robinson, a natural lefty who played baseball as a righty. He certainly goes against JR's theory, unless you hold he would have done better playing as a lefty.

However, the other variables, such as ballparks, eras, and all, change with time. Handedness does not. It is an integral quality, that for all intents and purposes cannot be changed. Win Shares, or most other measures for that matter, is not designed to measure that, because in the long run it doesn't matter. There isn't anything to be done about it anyway.

And yeah, we're getting off the Win Shares path into something else. If we want to take up the handedness issue, it should be in a different thread.

[gyb13]

I still haven't gotten around to reading WS, so I'll throw this out to those of you who have read it:

comparing the RSAA and WS lists in this thread (http://www.netshrine.com/vbulletin2/showthread.php?t=11439), i have to ask - does win shares have a bias in favor of closers over other relief pitchers?

[SmedIndy]

Yes, they do. Saves enters into the computation of win shares.

[Craig S.]

Yes, they do. Saves enters into the computation of win shares.

What I'm wondering is whether this aspect of WS is fair. If a middle reliever is as effective as a closer, should he be valued less because he doesn't accumulate a certain stat? Or does the closer deserve this because he has a tougher job?

[SmedIndy]

I think its used because holds and other advanced metrics aren't available for pitchers in the past. He's using what he's got for everyone.

[sweaver]

James' rationale is that the last outs are tougher, so it gives extra credit to guys what deserve it. I think he's off base on this one.

[Craig S.]

I noticed quite a while on Amazon that the book was no longer available, not even a used copy. Now, in the past few months, I've noticed 2 used copies for sale - one was at around $150 (it sold) and now there's one available for $250.

Was this title extremely limited in its production? Why is the price of a used copy so high?

[sweaver]

It wasn't a real big press run, and I'm sure people like me that have one would rather keep it.

[Craig S.]

It wasn't a real big press run, and I'm sure people like me that have one would rather keep it.

I've got no plans on selling mine, either. I guess it just surprised me that it was so expensive, so soon.

[Crash Course]

This could be because it was published by STATS Inc - and then once FOX took them over, they stopped doing books. Whatever was printed then just might be it.

I'm stupid. It would have been nice to see this coming and then buy a dozen copies when it came out.

[gigantes pablo]

Anyone looking for a copy of Win Shares might try this website for Stats, Inc.: http://www.stats.com/store/store.asp?page=wins.

I absolutely love this book. For all its many flaws, the Win Shares system and the book's format are incredible for getting a quick assessment of relative performance across years/eras. I also find the year-by-year list of win shares for each team to be incredibly informative.

Sorry to hear that the book's no longer for sale at Amazon. I hope this site and Stats' inventory of the book is current.

[Endymion]

Stats also sells a digital edition of the buck for MUCH cheaper than $250.

[Joseph]

Anyone looking for a copy of Win Shares might try this website for Stats, Inc.: http://www.stats.com/store/store.asp?page=wins.

I absolutely love this book. For all its many flaws, the Win Shares system and the book's format are incredible for getting a quick assessment of relative performance across years/eras. I also find the year-by-year list of win shares for each team to be incredibly informative.

Sorry to hear that the book's no longer for sale at Amazon. I hope this site and Stats' inventory of the book is current.

I'm a little confused here, is the link provided for the digital version? Can I order the paper version from stats inc?

[Endymion]

I'm a little confused here, is the link provided for the digital version? Can I order the paper version from stats inc?

Sorry for the confusion. Not sure if Stats is selling the paper version anymore, you can give it a try at the link above (http://www.stats.com/store/store.asp?page=wins).

The digital editions are available at

http://www.stats.com/store/store.asp?page=EWNS

They have the digital edition, which is a PDF(electronic book format) of the book and something called the Digital Update, which is career Win Share information for every player and some other stuff.

Hope that helps.

[Craig S.]

I called Stats earlier today, and they said they still have copies of the print version. I'm wondering how it can be the same edition that is selling for $200 on Amazon!