Saturday, February 2, 2013

Reality v. WAR

Perhaps my biggest problem with WAR -- the acronym of this stat being "Wins Above Replacement" -- is how a "win" should be defined.  I define a win by the results seen in the final score of the game, and some people (you know, me) would be smug and call this observation "fucking reality".  But this smugness is a reaction to the heightened douchery of the Sabermetic statheads who invented WAR, because it's impossible to figure out quite what the fantasy scenario is around their definition of a "win".

Let's use our favorite team ever, the wildly successful 2012 Boston Red Sox, as an example.  In the fantasy world of fWAR, this team went 116 - 46.  Surely, any time with such a win/loss record will roll over the feeble competition in the playoffs on their way to a World Series title (unless they are from Seattle).

Any team whose players amassed a collective WAR of 34.7 must have had over a hundred wins, right?  I mean, I just pulled up the stats of every player that swung a bat and threw a pitch for the Red Sox last season, added up all of the positive WARs and subtracted that by the negative WARs, and the total WAR for the 2012 Boston Red Sox is 34.7.

A Major League Baseball season is 162 games, so -- in the view of fucking reality -- a team with an 81-81 record would be .500, completely neutral, and have a collective WAR of 0.  Zero wins above, zero wins below .500.  81-81.  Therefore, a team whose players collectively batted and pitched for a WAR of 34.7 would have 116 wins, when you add 34.7 wins to the neutral total of 81 an round up.  And a team with 116 wins can have only lost a maximum of 46 games.

But, in fucking reality, the Red Sox went 69 - 93 last season.  So what happened?

What is a "win" in "Wins Above Replacement"?  Because, from what I see, WAR looks like "wins" pulled directly from the ass of Sabermetric statheads; and it doesn't equate with the reality of what happens on a baseball diamond.

Pretending that we can figure out the weight of a "win" in WAR compared with a win in reality, a WAR "win" for the 2012 Red Sox is worth about 0.577 of a win in reality if you add the players' collective WAR (34.7) to the difference of neutral/WAR 0 wins (81) by the amount of games the Red Sox actually won (69), the divide that number (46.7) by Neutral/WAR 0 wins (81).

But I'm willing to bet that a WAR "win" = 0.577 actual wins is not a constant in the formula for WAR.  This is just what a WAR "win" was worth to actual wins on the 2012 Red Sox, and a comparison of WAR "wins" to actual wins for other teams would return a different number for each and every team.

So, in what fantasy scenario is a "win" in WAR equatable with a win in reality, that shows up in a team's win/loss record?  I can't figure this out, largely because this value probably doesn't exist.  I can't find evidence that, when a player's WAR is calculated, that it pegs WAR to actual, tangible, victories.  This, essentially, reduces WAR to Sabermetric wankery that isn't really worthy of discussion.

In fact, it shows an inherent dishonesty among Sabermetricians that tends to piss off us old time dinosaurs who still think that stats like a player's batting average have a lot of value.  If Sabermetricians chose to call their cardinal metric something different, let's go basic with "Player Rating", that would be more honest.  Going further, if they provided a chart of how different variables were valued and weighed to calculate a player's rating, then we could have a discussion on the merits and the value of the stat.  I mean, we all know how a player's Slugging Percentage is calculated.  If you don't think a triple should be worth 3 points since triples are more an indicator of a player's speed rather than slugging ability, then you're free to look at other stats.  But I don't know how a triple is weighed in WAR -- I don't think anyone does.

Instead of being open and honest, Sabermetricians decided to speak in definitive terms and call their quintessential stat "Wins Above Replacement".  But they forgot to make "wins" a static variable, which would be very easy to do since wins exist in fucking reality.  That's why this stat is a joke.

No comments:

Post a Comment