THE BOOK--Playing The Percentages In Baseball

I’m not a BJO subscriber either, and so without any sort of context (I don’t even know what this is talking about - is it a revised Win Shares?) I can’t comment on the substance.

For someone that is a BJO subscriber, can you tell us what the Aparicio/Mantle issue is? And how many players it affects?

The reader pointed me to these articles:

http://www.billjamesonline.net/ArticleContent.aspx?AID=1073
http://www.billjamesonline.net/ArticleContent.aspx?AID=1075
http://www.billjamesonline.net/ArticleContent.aspx?AID=1077
http://www.billjamesonline.net/ArticleContent.aspx?AID=1080
http://www.billjamesonline.net/ArticleContent.aspx?AID=1082

If someone wants to quote from these, feel free.

He is attempting to answer a question that is really impossible to answer. Never the less it is an interesting question.  And, that question is “on paper, what was a team’s expected talent level for a given season?” To answer this he has a new method, of which he admits has many shortcomings. He claims that the method only needs to use broad strokes to get to second level questions…such as.. What is the most under/over achieveing team off all time? What is the most talented team of all time? How do managers perform vs. expected talent level?etc.

The basis of his new system is 1)age 2) career games player/innings pitched. With further bonus points for making the HOF or being MVP. With pitchers he uses career won/lost records. Catchers/relievers get different adjustments to bring them into line with everyone else.

What is dumbfounding to many of us is why didn’t he use win shares. He claims to not want to a players’s season stats because “ that leads us back to the question of how good teams actually were”. Many have asked him why not use the years weighted around the season in question?

He admits “ in some cases our estimates of a player’s expected performance are not as accurate as they could be” . The readers are lead to comment on a number of huge errors. Not the least of which is giving Aparicio’59 an expected value equal to Mantle’61. Then going on to say that the 222-216 edge the ’59 White Sox get over the ’61 Yankees is meaningful. BTW, I should mention that the values are similar to Win Shares, both Aparicio and Mantle score at 36. Further he goes on say that Red Ruffiing’31 had a much higher expected value than Lefty Grove in 1931, Nellie Fox’59 is higher than Gehrig ’31, Rollie Fingers ’73 is a 40 and Ruth’31 is a 29.

When it was pointed out that the actual WS/LS values for Mantle was 48-(14) and Aparicio 19-17.  He seems to get very testy with anyone suggesting an alternative approach

James states:

In constructing our scaffolding we have represented Mickey Mantle as being of the same size as Luis Aparicio, which of course is not exactly drawn to scale, and of course there are people who get hung up on that issue.  We have a few people on the site who want to stand on the ground floor and shout, “LUIS APARICIO WAS NOT AS GOOD AS MICKEY MANTLE.”
Well, yes, of course he was not.  What you are missing is, we are not discussing that issue.  We are not comparing Luis Aparicio to Mickey Mantle.  We are in a different room, talking about a different subject.  You can join us or not; that’s up to you.  You can stand on the ground floor and complain about the scaffolding being rickety from now until September if that’s what you want to do.  There are always people who do that; you’ll have plenty of company. 
At some risk of being rude. . ..well, it’s not exactly a risk of being rude.  I’m pretty sure this is rude.  My apologies.  To those of you who are hung up on the Aparicio/Mantle type of issues, let me point out what you are missing.  What you are missing is, everybody already knows that.  What you are pointing out to us is deafeningly obvious, and everybody except you has already got it.  I mean, I spent pages and pages in the opening article talking about the failings and limitations of my system, pointing out case after case in which my system failed for one reason or another.  What exactly did you think was the point of that?
Don’t answer that.  The point of that was to get past those kind of arguments, so that we could move on to the second room, and begin to survey that set of questions.  Yes, I understand that it’s not a perfect system.  It’s a temporary scaffolding.  We have to start somewhere.

What everyone is wondering is if Mantle’s WS were a 30,36,33,14 around ’61 and Aparicio was a 19,20,13,18 around ’59. Why Mantle (with recent season’s weighted 2x) wouldn’t be a 30 and Aparicio as 17. Thus completely flipping his analysis of comparing 59 Sox to 61 yankees. Why didn’t he use Win Shares which he once described as designing to answer these types of derivative questions? Why does anybody who questions or offers an alternative, infered to be dumb?

While I myself often question BJO’s value compared to everything else that’s freely available (like this site), I can’t believe that Tango has bailed on James.

Anyways, as a BJO subscriber, and perhaps someone who has probably failed to recognize he’s being ripped off, I’ll provide some feedback.

Last week James introduced his new “Teams on Paper” method in a 4 part series.  The method attempts to take a retrospective look at the players that made up a team in any given year and figure out how good that team should have been.  Basically he’s looking at a players career in hindsight and then applying an aging factor to assign each player’s season a point value.  He does not look at the actual outcome of the player’s season in question, but at his career in it’s totality.  It’s sort of a retrospective Value Approximation Method.  In the end he totals up the team’s points and estimates how many wins they should have had.  He’s using it to see what should have been the greatest / worst teams of all time, who were the biggest over/under achieving teams.  It’s a decent read.

Anyways, as part of the process he did a comparison of the ‘59 White Sox and the ‘61 Yankees.  He gives a 25 year old Apparicio and a 29 year old Mantle both 36 points.

One of the reader’s jumps on this and says… “Mr. James you are a genius, except when you publish systems that say that Luis Aparicio was as good as Mickey Mantle”.

A few days later, in James closing piece, he spends three paragraphs beating up on this guy’s comment.  Well, you can see the reader’s point, which suggests that the system may have a few holes.  James more or less admits this, but does seem to go out of his way in attacking the reader.

However, as a long time reader of James, it didn’t startle me to see him take to the offensive against a critic, albeit a rather small blip on the radar.

I don’t see how we can be expected to join James in the next room if he leaves an issue like that hanging.  I don’t subscribe anymore either so I won’t see his explanation.

The best way to do this would probably be to use a projection system, but use the years on both sides of the year in question.  Not sure how you’d handle injuries - say a great player misses a whole year with an injury, but is great for 700 PA the 2 years before and after that lost season.  Would you consider this a team that plays under it’s potential or are we asking a different question?

How about if the same player wasn’t hurt, but only got 150 PA because the manager got mad at him for some reason?

ksc/3:

James once used Approximate Value to compare the Big Reds Machine to (I think) the 90s Yankees.  He did something like weight the current year double, and then the surrounding years less, and he used Approximate Value. 

It sounds from the description that this is what James wants to do, but he’s doing it with less granular data.  Almost like saying “this is a smooth parabola of Cal Ripken’s career, and regardless of how much above or below his PERFORMANCE is of that line, we will treat his TALENT as exactly on that line”.  I don’t blame him if he is taking that view.  It’s a decent view to have.

From the description however, he is taking that view, but doing even less work in trying to find that line for each player.

That said, if you want it done a (much) better way, then look to MGL’s articles from the last two years in Hardball Times Annual, and Phil Birnbaum’s research presentation at SABR a few years ago:
http://www.philbirnbaum.com/presentation.zip

So, in answer to James’ question about starting somewhere, I’d say to start where some of the great work has left off.

***

Wes: I bailed because I didn’t think I was getting anything out of it any more.

I don’t see how what James is trying to do is much of an issue.  It is pretty trivial as Rally implies.

You simply do a projection on a player for that year using data from before and after that year AND the data from that year.  Obviously you do an age adjustment like you do with any projection. If you know that a player was injured for that year, you can adjust for that, or more heavily weight that year (the former is better, as a player could have been injured AND gotten lucky, such that his stats for that year don’t reflect his “injury laden true talent"). 

So what am I missing here?  It doesn’t seem to be hard to do this. Why does BJ have to invent a new system?

This does not appear to be James’ best work.  I’ll resubscribe when he decides to offer his best work to his readers, instead of saving anything really worthwhile for the Red Sox.

Not that I blame him, I’m sure the Red Sox pay a lot more than he can make from a few hundred people paying $3 a month, but I only have so much time to spend reading baseball stuff online, and while I was a subscriber I visited the site about once a month on average.

It certainly wasn’t a daily draw like THT, Fangraphs, this site, or BTF.

To me. James is like the crazy, genius inventor that has thousands of patents and likes to tinker around with all kinds of things, some interesting, some not, some good, some bad, some important, some just for fun, etc.  Every once in a while they come up with something really important and profound.  Lot of times they come up with something profound, but it is just for fun.  That is Bill James, to me.  I just read his new Gold Mine.  Again, all kinds of stuff in there, but almost nothing much interesting to me.  There is an interesting article on batter fly balls and ground balls, but that is about it.  I have not read the whole thing yet cover to cover though.  He seems to be tinkering around all the time with “toys.” I don’t know how he would have the time to do serious work for the Sox.  Maybe he is just a figure head for them at this point - I don’t know.

It should also be noted that BJ believes he is the first one to examine this issue. Phil Birnbaum has done a magnificent presentation, that includes all the real data, as Tango has noted.

Which SABR convention did Birnbaum do that presentation at?  IIRC, James may have actually been in the audience (he was at the Toronto convention.)

Phil’s home page says:

“My Presentation from SABR 35 in Toronto,”