November 16, 2005
Can Of Corn
The Case for Pettitte
My frustrations during awards season are serialized, perennial and enduring. This year, of course, is no exception. I could rail about the risible snubbing of Johan Santana in the AL Cy Young balloting or the fact that Jason Bay, the third-best player in the NL, finished a paltry 12th in the NL MVP vote.
However, my prevailing gripe is that Andy Pettitte received short shrift in the chase for the NL Cy Young from credentialed voters and statheads alike. There's a reason for this, and it touches on a larger issue that's of much import. That issue is this: the way we employ park factors in the analytical community is wrongheaded, and it needs to change posthaste.
As you're aware, we mostly use runs-based park factors when making thumbnail or even quantitative adjustments to statistics. This is problematic on a couple of fronts. First, doing so ignores how parks affect run scoring on the component level. Knowing that a park has a runs factor of 103 tells us that it promotes the scoring of runs, but it doesn't tell us whether it does by inflating home runs, singles, non-home run extra-base hits or all of the above.
Additionally--and this is a more vital failing of runs-based park factors--they don't tell us how a park accommodates left-handed batters and right-handed batters. For instance, Comerica, Wrigley, Safeco, Fenway, Shea and many others affect hitters quite differently according to their handedness. To cite one extreme example, from 2002-2004, Minute Maid Park's average home run factor for right-handed batters is 116, while its home run factor for lefty hitters is 84. That's a profound disparity, and applying one blanket factor to all hitters isn't sound analysis. We have the data available to do better, and do better we should. The days of using runs-based park factors or park factors not broken down into platoon splits should be over.
In any event, the aforementioned Minute Maid Park example provides a tidy jumping-off point into a discussion of Pettitte's merits. First let's look at how the NL Cy Young contenders compare in a handful of key measures:
IP RA PK_RA VORP Carpenter 241.2 3.05 3.08 68.4 Clemens 211.1 2.17 2.17 80.6 Pettitte 222.1 2.67 2.67 72.4 Willis 236.1 3.01 3.20 68.1Lay aside any misguided affinity for pitcher win/loss records, and the serious discussion comes down to Clemens against Pettitte. Clemens holds the edge in RA, park-adjusted RA and VORP. However, VORP and PK_RA are park-adjusted using the broad strokes of the runs-based park factor. According to these and most other park-adjusted metrics, Clemens and Pettitte toiled in the same environment this past season. That's demonstrably false. To suggest that the right-handed Clemens and the left-handed Pettitte are equally prone to the foibles of Minute Maid is delusive. Here are the platoon park factors for Minute Maid in 2005 and from 2002-2004:
LHB AVG LHB HR RHB AVG RHB HR 2005 94 86 104 140 2002-04 100 84 104 116As you can see, Minute Maid is a haven for right-handed batters, but it's thoroughly hostile toward lefty hitters. This phenomenon, suffice it to say, has direct bearing on how we should view the accomplishments of Clemens and Pettitte this past season.
In 2005, Clemens worked 54.7% of his total innings at home, while Pettitte logged 53.4% of his total workload at Minute Maid. That's roughly equal, so now we need to assess the "Minute Maid" effect. Here's what percentage of lefties and righties each pitcher faced at home:
%LHBs %RHBs Clemens 50.7 49.3 Pettitte 22.8 77.2Overall, Pettitte faced right-handers at Minute Maid in 41.2% of his total 2005 plate appearances. As you can see above, those are highly hostile conditions, and that's obscured in any metrics that involve runs-based park factors. When you account for the fact that Clemens faced left-handed batters at home in roughly a quarter of his 2005 innings (and thus benefited from those circumstances), and that more than 40% of Pettitte's opposing batters were right-handers in Minute Maid, it's clear Pettitte had a much tougher go of it than Clemens did.
So Pettitte threw more innings than Clemens and pitched in a substantially tougher environment than did his teammate. That's why, in my opinion, Pettitte should've been the NL Cy Young winner in 2005.
Even if you disagree on that count, the larger point is that component and especially platoon park factors need to become a part of the stathead SOP going forward. The aim of recalibrating for environment is an eminently sensible one, but surface-level adjustments that gloss over the nuances of how parks influence the game on the field should no longer be acceptable. Doing so not only deprives isolated performances (like Pettitte's in 2005) of the credit they merit, but it also compromises our understanding of the game at large. Let's do better.