Win Shares?

by Uncle Charlie on Jan 27, 2005 10:04 AM PST

I was reading through the Bill James Handbook last night and I started reading the Win Shares section and was a little confused by something. Maybe some of you fellow AN folks can help me out.

James says that Win Shares is a way of calculating the overall contribution that a player makes to his team -- offensive, defensive, and pitching. Fine. Here's the confusing part. He says that a player's win shares will not be affected by the quality of his team -- that a great player on a bad team will have as many win shares as a great player on a good team. So, to me, that sounds like everyone is essentially normalized and treated as thought they were on an equally-good team.

Then, James goes on to say that if you take the number of games a team wins in a season and multiply it by three, you will get the sum of the win shares for all of the players on that team in that season. So, for example, the A's won 91 games last year -- we multiply that by 3 and we get 273; if we take the sum of the win shares of all of the players on the A's last year, the total will be 273.

So, what I don't understand is, how can it be that a player's win shares aren't affected by the quality of the team he's on? If Jason Kendall is on the Pirates and they win 72 games, isn't the number of win shares that can be allocated to Kendall more limited than if he is, say, on the A's. The A's won 19 more games last year, meaning that there were an extra 57 win shares to be distributed among their players. How can it be that the quality of a player's team does not affect the number of win shares that are allocated to him?

Tweet Comment 8 comments | 0 recs |

Do you like this story?

Comments

In theory ...

It works out that if a team creates 21% more win shares (as the A's did, compared to the 'rats), they should have, more or less, 21% more talent to divide the win shares among. So a 30 win share player on one team is still a 30 win share player on the other, but in one case, he makes up ~11% of the win shares and, in the other, he makes up ~14%.

There are many faults with this system, though. The biggest is that it assumes that offense, defense, and pitching each count for exactly 1/3 of a win. In reality it should be more like 45% O, 15% D, 40% P - on a balanced team. Which brings us to the other problem.

Many teams aren't balanced. Many teams are much more reliant on either offense, defense, or pitching than is typical. The A's or Dodgers of 2003, for example. The A's were almost completely reliant on their pitching to carry the team, while the Dodgers relied on their defense more than is typical. In cases like this, the team won more games than the offense earned them. Regardless, offense still counted for the exact same amount as pitching and as defense - so a hitter on a team like this would get the proper proportion, in relation to his contributions to the offense, but of a pie that is larger than he deserves - leaving him with more offensive win shares than he deserves. The converse is true, that a pitcher on the A's or a fielder on the Dodgers would receive fewer win shares than they deserve (well, the Dodger fielder would still get more than any fielder deserves, because the stat overvalues defense, but, relative to other players [who are similarly overvalued] less than he deserves).

Relax alright don't try to strike everyone out. Strikeouts are boring besides that they're fascist. Throw some ground balls it's more democratic.

by devo on Jan 27, 2005 10:27 AM PST reply actions

Never mind ...

I'm wrong - Oaktoon is right - I think I'm describing an older model.

Relax alright don't try to strike everyone out. Strikeouts are boring besides that they're fascist. Throw some ground balls it's more democratic.

by devo on Jan 27, 2005 10:34 AM PST up reply actions

I'm sure I didn't just make that up, though

Do you know what I was talking about? [Since I clearly don't ; )]

Relax alright don't try to strike everyone out. Strikeouts are boring besides that they're fascist. Throw some ground balls it's more democratic.

by devo on Jan 27, 2005 11:02 AM PST up reply actions

I don't understand your argument

for an unbalance assignment for O, D, and P. One argument that I learned is that at each instance of the game, each component of O, D, and P would contribute to similar amount of out's and bases. Since Win's is proportional to outs and bases, James made the equal assignment to each component of the game.

The fact that baseball players ought to make the routine defensive plays doesn't mean that that play contribute to the win any less than a single does or a strikout. It just means that in regarding to defense win share, the range of the worst to the best is not as large as the range in regarding to Pitching and Offense.

That is why early on (late 90's), BB doesn't care too much about defense. 1 extra defensive win share costs much more than 1 extra pitching win share or 1 extra offensive win share at the Economics of that time because 1 extra defensive win share would take someone like Ellis into a player that shows up on ESPN gems once every other day or so.

What a pitch!

by monteverdi on Jan 27, 2005 11:40 AM PST up reply actions

I guess I got to go back to read James argument

more carefully to see what he makes about unbalance assignment.

What a pitch!

by monteverdi on Jan 27, 2005 11:41 AM PST up reply actions

Did you even read the rest of the thread?

First of all, you need to group creating bases with avoiding outs (offense) and creating outs with avoiding bases (defense/pitching).

Once you do that, it's clear that offense should contribute roughly 1/2 off total winshares.

Relax alright don't try to strike everyone out. Strikeouts are boring besides that they're fascist. Throw some ground balls it's more democratic.

by devo on Jan 27, 2005 11:48 AM PST up reply actions

Well

Every team does not record the same number of outs. When you consider extra inning games, games not played because of rain and not rescheduled because of standings, teams losing on the road not getting three outs for the bottom of the 9th, ect, there is plenty of room for variation among teams.

by RichardP on Jan 27, 2005 11:05 AM PST reply actions