Printer-friendly

July 26, 2005

Can Of Corn

Putting the Park Back in Park Factors

by Dayn Perry

The (excruciatingly obvious) recognition that environment affects the game on the field has made impressive inroads in recent years. It's not uncommon these days to hear rank-and-file fans or mainstream analysts paying qualitative heed to this notion, which is a good thing. In seamhead circles, most commonly this discussion takes the form of park effects, which, as you know, entails making statistical calibrations to reflect the tendencies of a particular ballpark. After all, a run scored in Dodger Stadium in 1968 isn't the same as one scored in the Baker Bowl in 1932 or Coors Field circa 1998.

Speaking of Coors Field, thanks to its incontrovertible effects, you'd be strained to find any writer or team exec not at least dimly aware of how playing environment influences the game. For too long however, we've been reliant upon the broad brushstrokes of runs-based park factors; i.e., evaluating a park by its gross effects on runs scored while being heedless of its effects on component-level statistics like AVG, OBP and SLG. For instance, Dodger Stadium, generally speaking, suppresses run scoring not by cutting down on homers, but rather by cutting down on almost every other positive offensive event. Recognizing this conveys much more "actionable" knowledge than lamely offering, "It's difficult to score runs at Chavez Ravine."

Beyond even component-level factors is the notion that batters will fare differently in a certain park based on their handedness. For instance, Comerica cuts down on home run rates overall, but it does so by narrowly hindering left-handed power hitters and altogether crippling those hitting starboard. It's also worth pointing out that looking at single-year park factors is, I think, improper. At least a three-year factor should be used if you're interested in drawing any meaningful conclusions. All of this is to say the old method of looking at a single season's worth of runs-based park factors is wholly inadequate. You need multiple years' worth of data, and you need not only component-level factors, but also component-level factors broken down into left-handed and right-handed splits.

This raises the subject of platooning. For too long, platoon decisions have been informed by the handedness of the pitcher and the handedness of the hitter. Those are certainly eminently sensible considerations, but, in light of what we know about the indigenous tendencies of certain types of players, ending the platoon discussion at that point is darn well uninspired. Groundball and flyball tendencies of batter and pitcher should be considered, as should the micro-level dispositions of the park in question. This latter point can be particularly informative insofar as it applies to home runs, which, in my opinion, best combines variance and sample size among component stats. By that I mean, unlike, say, triples, you have enough home runs hit in each park to draw quasi-firm conclusions, and there's also enough variation from park to park that teams can use the information to gain an advantage. So should teams be wielding "park platoons"? Yes, in certain circumstances. Using 2002-2004 park data, observe how the following parks affect the HR rates of left-handed and right-handed batters:

Park               LHB HR Factor          RHB HR Factor
Astros             84                     116
A's                97                     114
Cubs               97                     137
Devil Rays         98                     75
Mariners           113                    86
Mets               115                    77
Red Sox            76                     105
Rockies            147                    127
Tigers             93                     74
White Sox          121                    137