THE BOOK--Playing The Percentages In Baseball

Thursday, May 07, 2009

Yes, the relationship between WAR and salary is linear

Here’s the chart from Jeremy:

Whether linear or exponential, the correlation (r) is .79-.80, in either case.  So, really, is it worth it to make it exponential?  I know it’s not worth my time.

And one of the commenters is correct that you want to look at forecasts, not prior WAR.  No big deal though in this case.

Jeremy also looks at the (old) WARP data, and he has it exponential.  There’s a good reason for that: the extra-low replacement level forces the issue.  I have no doubt that when the new WARP makes its official debut that the new MORP will look unsurprisingly like my WAR to $ model (as best seen at Fangraphs).