December 08, 2008
UZR/PMR Comparison
Dan Turkenkopf runs correlations of UZR and PMR and I find them surprisingly low, given the two systems use the same data.
Both PMR and UZR were calculated using the Baseball Info Solutions (BIS) data set this season. I wonder if David or MGL might be able to give some ideas as to where the differences might come from.
I don't know enough about the UZR calculations to speculate. I base my models mostly on visiting players in parks, however. UZR might use all the data. I also don't know if UZR, like +/-, doesn't penalize players for outs made by others. In PMR. If the right fielder catches a ball that the centerfielder might be able to catch, the centerfielder is penalized. In +/-, the centerfielder is not. Given the low correlation with centerfielders, I suspect that's the case.