With that in mind, I went back to fangraphs and sorted by all the players who have played at least 250 innings in a season in RF and in LF since 2002 -- the year that UZRs began to be recorded on the site. I downloaded them to excel -- there were a little more than 700 RF and LF seasons of more than 250 innings -- and sorted by player. It seems to me that, if the positional adjustments for RF and LF are equal, then someone like Juan Rivera -- who's played more than 2000 innings in LF and more than 700 innings in RF over this period -- would be an equal defender in both spots. In other words, his UZR/150 should be equal for RF and LF. So I used only the players who had more than 250 innings in a season in RF AND 250 innings in LF over that time period. People like Ramirez and Dunn were deleted b/c they've only played in LF, for example. I then sorted by position, figuring that, if the positional adjustments are truly equal, the UZR/150 for each position would be roughly equal.
There were 165 seasons from 2002 to today where the LF played at least 250 innings and 167 seasons where the RF played at least 250 innings. These LFs played 92,702 innings since 2002 and the RFs played 104,402 innings in RF. The LFs accounted for a total UZR of 138 runs above average, for a UZR/150 of 2.002. The RFs -- we're talking about the exact same players, remember -- accounted for a total UZR of 22.8 runs above average for a UZR/150 of 0.295. Now, I didn't run a regression b/c I lack the software (and have forgotten how to do it anyway!) but to me a difference of 1.7 runs of so when we're talking about 100,000 innings strikes me as being statistically significant (though I absolutely cannot confirm that). If it is, shouldn't the positional adjustment for left field be 1.5 - 2 runs lower than it is for right field?
Doesn't it make sense that fielders deemed to be above average in LF are compared to worse defenders than those in RF, thereby punishing RFs by the fact that they don't get to be compared w/ Dunn and Lee? Would Holliday still be 6.6 runs above average if he had played in RF since 2004? Are the positional adjustments just approximations such that this 1.7 run difference doesn't matter all that much in the grand scheme of things? I'm looking for help on understanding the positional adjustments a little better b/c God knows that Tango, MGL, and Dolphin know a hell of a lot more than I do.
This link should, if I've done it right, take you to the google spreadsheet so that you can look at my data. All I've done is download from fangraphs, sort them, delete those who didn't fit the criteria, sort again, and create 2 sums and figure UZR/150.