from Amazon
Friday, December 04, 2009
Is the WAR-salary model broken?
In light of the Polanco / Scutaro signings among others, someone asked:
4) Are we sure the “model” is even right?
The model is right. The model says that wins are paid for at a linear rate, above the replacement level.
The model, being a model, and not a forecaster, would need to be fitted after-the-fact. If it turns out that the dollars per win is 3MM this offseason, then that’s the model. We’ll know soon enough.
So far, all the middle infielders signed in the off-season followed a pretty set pattern: Polanco, Scutaro, Wilson, Gonzalez, McDonald. Rank them by expected WAR, and that’s the list you’d get. Rank them by salary, and that’s the list you get. Rank them by salary/WAR, and they’d probably be equal. It works.
If it turns out that the middle-tier is paid 3MM and the elites are paid 5MM, then the model is broken. For example, if Holliday signs for 6/150, the model is broken. If he signs for 6/100, the model is fine.
(Note the additional provision that the model doesn’t try to best-fit it to individual stats like HR, RBI, and W. Those are biases. If the model wanted to include those as parameters, it could. Luckily, there’s no guys like that to scr-w up the model. No Carlos Lee or Sorianos or Howards.)