Table of Contents

Monday, November 12, 2007

PMR on the Reds

David Pinto continues to release his PMR data at a steady clip. I consider PMR superior to RZR because of the additional adjustments it makes for batter handedness and park factors, so I will be updating my composite fielding estimates and total value estimates for Cincinnati Reds based on his data once he's run through the seven non-catcher/pitcher positions.

For now, though, it's nice to see some Reds getting love from Pinto's system:
  • Brandon Phillips - +46 outs, or ~+35 runs. Wow.
  • Josh Hamilton - +4 outs (+3.6 runs) in CF. This will push his overall fielding estimate in CF above average in 2007. The Fans also rate him as just above average, so I believe it. Maybe he'll be ok to have out there, at least for the next few years.
  • Norris Hopper - +1 outs in CF.
  • Alex Gonzalez - +0 outs at SS. That's a disappointment, though he'll still be above-average based on the other stats.
  • Jeff Keppinger - -6 outs at SS (-4.5 runs).
Update: Edwin Encarnacion was rated just -7 plays below average (-5.6 runs) at 3B according to PMR, which is much better than his -21.9 RZR rating this season, and much more consistent with his Fans' and ZR ratings. Why? I'm not sure, as PMR and RZR use the same hit location data; maybe there were an unusually large number of lot of hard-hit balls through his position this year? Replacing Eddie's PMR value with his RZR value will give Eddie's total value a boost of ~5 runs or so, which will shoot him into Scott Hatteberg territory. Very nice to see some more numbers supporting the conventional wisdom that he improved significantly this season with his glove.

Also interesting are his estimates of fielding behind pitchers, which can be considered a more precise way of assessing BABIP (or DER, depending on your preference) because it considers hit location, batted ball type, etc:
  • Aaron Harang, +5 outs
  • Matt Belisle, -0.5 outs
  • Kyle Lohse (when with Reds), -3.7 outs
  • Bronson Arroyo, -5.6 outs
If we convert those out values to an approximate runs stat (I'll use 0.8 runs/missed out), we can then subtract those values from actual runs allowed and get an approximate fielding-independent R/9 (listed below as PMR-R/9). I'll list that along with FIP, another way of trying to normalize pitcher performance:
Name ERA FIP R/9 PMR-R/9
Harang 3.73 3.70 3.88 4.04
Belisle 5.32 4.54 5.62 5.60
Lohse 4.58 4.50 5.19 4.99
Arroyo 4.23 4.51 4.66 4.47

FIP is an estimate of ERA based strictly on readily available pitcher peripherals -- k/9, bb/9, and hr/9. Differences between FIP and ERA, in general, are due to poor fielding, "clumping" of offensive events in a manner that deviates from the norm, or non-hr park effects. Differences between R/9 and PMR-R/9, on the other hand, can probably be considered to be strictly due to differences in fielding.

Generally spreaking, the differences between FIP and ERA are much larger than differences between R/9 and PMR-R/9, at least among these four pitchers. In fact, differences between the two pairs of stats only move in the same direction twice! This indicates that the timing of offensive events (which is largely the result of luck) may have more to do with differences between actual and DIPS-based ERA estimates than the Reds' below-average fielding. That jives with Tom Tango et al.'s breakdown of variation in BABIP:
luck: 44%
pitcher: 28%
fielding: 17%
park: 11%
Neat stuff.