Sunday, December 10, 2006

Projecting Pettitte

Now that Andy Pettitte is a Yankee again, the major question becomes, how will he adjust to the AL? There is a very obvious imbalance between the leagues, and surely Pettitte will suffer for it. But how much?

There were 40 pitchers who split 2006 between the AL and NL, totaling over 2800 IP. I put their stats in both leagues side-by-side and weighted each player by the least PA. For example, Shawn Chacon faced 303 batters with the Yankees and 206 batters with the Pirates. I therefore scaled back his AL stats to 206 batters faced so that he affects each league equally. Here is how those 40 pitchers performed in each league, on a per 200 inning basis:

LeagueIPHHRBBKERAFIPAVGOBPSLG
AL20023733801295.725.35.296.362.493
NL20022129771465.184.89.281.349.461

You can see that switching from the NL to AL will add half of a run to a pitcher's ERA. There is also an across-the-board shift in every statistic showing the NL as the 'easier' league. I don't think any of these numbers will surprise anyone.

Now, to apply these to Pettitte. First, we need a baseline projection. We'll simply use a Marcel-style system: a weighted average of the past three seasons. For Pettitte, we get the following line:

GSIPHHRBBKERAFIPAVGOBPSLG
3119018620551593.603.82.257.306.408

Pretty darn good. If Pettitte were returning to Houston, that's the line we'd be predicting. But we want to know what he'll do with the Yankees. First, we adjust the projection from the NL to the AL:

GSIPHHRBBKERAFIPAVGOBPSLG
3119020223571424.224.24.272.321.438

Doesn't look so hot anymore, does it? Keep in mind, this is the same exact performance as the previous pitching line; the only thing that's changed is the context.

Next step: park factors. Yankee Stadium is presumably an easier place to pitch for a southpaw than the Juice Box. Using the Day-By-Day Database, I figured component park factors for the past three seasons for both stadia. Doing this drops Pettitte's ERA to 4.10.

Last step: defense. As a lefthanded groundball pitcher, Pettitte relies rather heavily on his shortstop. Thankfully, he had Adam Everett sucking up everything in sight. According to The Hardball Times, the Astros made 86 more plays than average on groundballs (tops in the majors) but made 31 less plays on flyballs. Meanwhile, the Yankees were -9 on groundballs and +14 on flyballs. Applying the appropriate weights to Pettitte's projected number of grounders and flies, and we would expect him to give up an additional 10 hits with the Yankees.

I'm regressing that number, however. Since I only have the 2006 stats on this front, I would rather be conservative and bring them back to average 50%. So, our final, league-adjusted, park-adjusted, defense-adjusted, super complete 2007 projection for Andy Pettitte is:

GSIPHHRBBKERAFIPAVGOBPSLG
3119020522571424.274.12.277.325.439


That's what the Yankees are paying $16 million for. Meh.

And just for kicks, here's Roger Clemens's projection:

TeamGSIPHHRBBKERAFIPAVGOBPSLG
Astros2616312510471492.313.13.212.271.314
Yankees2616314011491332.853.45.231.290.337

1 comment:

Jeteupthemiddle said...

Pretty good, anf.

SG at Replacement Level Weblog, got a similar projection using ZiPS--an ERA of 4.30.

Interestingly, when he projected Igawa, he got an AL projected ERA of 3.97. http://yankeefan.blogspot.com/2006/11/projecting-kei-igawa.html

I hope/still think they will sign Igawa.

Which leads me to the conclusion they will trade Pavano for whatever comes along.