THE BOOK--Playing The Percentages In Baseball

Tuesday, January 18, 2011

James Player Rater 1994

Carson compiles the Bill James data to see how he did.  He was also nice enough to send me his file, on which I ran a regression of James’ dollar values and their eventual WAR.  The correlation was r=.60, which, I dunno, sounds high for something like this.

The regression equation is this:

expected WAR = 1.3 * James$ minus 16

Take for example James’ highest rated player: Cliff Floyd at 44$.  That would translate to an expected WAR of 41 (basically Jim Rice or Dave Parker).  Floyd got 27 WAR.  He also had Chipper at 40$, meaning 36 WAR, and he’s at 86 and counting. 

If you take James’ top 10 rated players, they averaged 36$, which means an expected 31 WAR for their career, and they got 39 WAR.

Working backwards, in order to estimate Chipper at 86 WAR, his James$ should have been 78$.

Anyway, just fascinating stuff, and I will make my plea for all forecasters to please release your historical data to us researchers.  Anything more than ten years old.  It’s useless to you, and a goldmine to us.