WAR - what is it good for?

[OttoinGrotto]

Nice article about WAR. Doesn't really convince me though, because I think WAR has a fatal flaw. Here's a quote:

In 2012 the correlation between Baseball Prospectus' WAR and team victories was 0.86 (where 1.0 would have meant a perfect correlation). The correlation between batting average and victories was 0.27. Teams with more WAR win more games.

And the linky - http://espn.go.com/mlb/story/_/id/8...-next-big-all-encompassing-stat-espn-magazine

The big issue that I have with WAR is that actual team performance, ie winning isn't part of the formula. A lot of the stats used to calculate WAR are proprietary, so there's not a way to check it and maybe there's a compelling statistical reason why winning's not a part of the formula, but holy crap it seems strange that park effects and quality of opponents are part of the formula, but the player's actual team performance is not. I don't view the correlation between WAR and team victories as being in support of WAR at all. It feels like it's pointing out a key missing variable.

 

[Knicks411]

Bill James created win shares like a decade ago and it was tied directly to team wins; it was 3 win shares per win, so if a team won 100 games, the players on the team would end up with 300 win shares no matter what. Most of the time, runs correlates pretty well to wins, so just working with runs will get you nearly all of the way there. And most of the time, when a team wins more (or less) games than their run differential says they should, it reverts the next year.

I like WAR. Obviously it's not perfect. (And it definitely can be frustrating when Fangraphs and BBref can have wildly different numbers for the same guy). But a correlation of about .86 seems right to me. If one player has 5 WAR for a season and another guy is 4.5, then I'm not going to be convinced the first guy is better, but if its 5 to 3.5 or something, then I would take that as pretty good evidence the 5 WAR player had a better season. (Or it it's 10 to 7.1, like it was in a certain MVP race, then I'm even more convinced)

The big issue to me, and I would assume most people, is the defensive component. I think for the most part we've gotten offense down pretty good. Baserunning not as much, but it's not that difficult to see how many times a guy takes the extra base, etc, and anyway, most of the time, the base running totals are pretty low anyway so it's not a huge deal.