Out With the Old UZR, in With the New

Boy, FanGraphs just keeps getting better and better. If you haven’t heard already, they’ve not only brought in some 2009 UZR data but they’ve revamped the metric a bit. Notable changes include improved park factors (generally speaking, the more data the better- and now there are more adjustments for the “quirky” aspects of parks, like Fenway, etc.) a soon to arrive home/road splits section, and positioning adjustments based on the strength of the hitter (for OF) or the speed of the hitter (for INF). Very cool stuff. With all these little (or perhaps not so little) adjustments made, how different is the new UZR from the “old” one?

If we look at “qualified” players only, from 2003-2009, we get:

r	r^2	MAE	Max	#10+%	#5+%
0.952	0.906	2.09	18.7	1.0%	7.5%

That’s a very strong correlation, and the average difference between the original UZR and the new one is around 2 runs. The maximum absolute difference was an astounding 18.7 runs, but occurrences such as this are exceptionally rare- of the 941 players sampled, only 1% of them showed 10 runs of difference (hence the #10+%) and 7.5% were 5 runs and above. This is only a simplistic look, however- let’s break it down by position and component. First, range (sans error runs; these will be shown separately):

Range	r	r^2	MAE	Max
1B	0.934	0.871	1.69	7.7
2B	0.951	0.905	1.75	7.6
3B	0.975	0.950	1.47	5.9
SS	0.953	0.909	1.91	8.6
LF	0.876	0.767	3.61	18.0
CF	0.954	0.911	2.34	9.6
RF	0.962	0.925	2.09	8.0

First base appears to show the largest difference among the infield positions in terms of r^2. Yankee fans will be pleased to know that Mark Teixeira was no longer a -3.7 defender in 2009- he’s now a +0.1- but the largest difference in estimated range goes to Albert Pujols among first-sackers, going from 13.7 to 21.4 runs. Second baseman Orlando Hudson (when with the Diamondbacks) was originally rated a -1.9; he’s now a -9.5. At shortstop, former Dodgers shortstop Cesar Izturis rated a -8.8 and has moved up to a -0.2, and former teammate Adrian Beltre sees a boost of +12.5 to +18.4 runs.

We’re looking at a more noticeable difference in the outfield, and especially in left field. I assume this is where the park adjustments are really beginning to come in to play. The player that saw the largest difference of value, however, was Moises Alou when he was with the Cubs- -4.6 runs to +13.4 (in 2003). He also saw a dramatic increase in 2004, going from +5.6 runs to +19.3. Jacoby Ellsbury saw a change of -16.1 runs to -6.5 in center last season, and Michael Cuddyer saw a boost from -17.3 runs to -9.3 (2009). For Ellsbury and Cuddyer, they were no longer horrible- merely well below average. Although I suppose you could say that it’s “putting lipstick on a pig.”

On to error runs:

Errors	r	r^2	MAE	Max
1B	0.852	0.727	0.80	2.4
2B	0.980	0.960	0.33	1.8
3B	0.991	0.982	0.37	1.6
SS	0.991	0.983	0.33	1.4
LF	0.999	0.998	0.02	0.2
CF	0.994	0.988	0.04	0.6
RF	0.997	0.993	0.04	0.7

While left field didn’t match up so well with range, it was essentially a perfect match for errors. First base sees the largest discrepancy, although it’s rather small. The biggest change goes to Derrek Lee of the Cubs, going from +1.1 runs to -1.3. All in all, we’re looking at a very little difference.

Arm/DP	r	r^2	MAE	Max
1B	0.969	0.940	0.08	0.4
2B	0.991	0.982	0.16	0.7
3B	0.995	0.990	0.04	0.3
SS	0.989	0.977	0.17	0.7
LF	0.979	0.959	0.68	2.0
CF	0.988	0.976	0.55	1.2
RF	0.980	0.961	0.85	2.0

Once again, we’re looking at essentially the same numbers, with very little difference between them.

If we put it all together, we wind up with:

UZR	r	r^2	MAE	Max
1B	0.939	0.882	1.76	6.4
2B	0.964	0.930	1.74	6.4
3B	0.978	0.957	1.49	6.2
SS	0.963	0.927	1.92	8.5
LF	0.885	0.782	3.61	18.7
CF	0.962	0.926	2.31	8.8
RF	0.969	0.939	2.20	8.6

All in all, we’re looking at about a 2-3 run difference. Nothing substantial, aside from the occasional outlier.

Those outliers, by the way, appear to reside in left field. All of the top 10 players in UZR difference play LF. 4 played in Boston (Manny Ramirez and Jason Bay), Matt Holliday (when with Colorado), and five Cubs (Alou, Alfonso Soriano and Matt Murton). So, it would certainly appear that left field is by far the most affected position.