Friday, November 14, 2003
Back! With Baseball blogging!: David Pinto has been experimenting with a probabilistic model of range (scroll down to the September 19th entry, and read upwards). In his most recent post, he lists all pitchers with more than 200 balls in play (BIP) and lists them by Expected Defensive Efficiency Rating (DER – see here for an explanation of DER, you need to scroll to the bottom.) WARNING: The next few paragraphs are heavy in statistics.
First things first. 109 pitchers have a negative DER Difference and 121 have a positive DER Difference, so we’re dealing with a sample of 230 pitchers. (Mean: .000557, Standard Deviation: .016902) Let’s define an outlier as two standard deviations outside the mean (95% confidence rate). The range we’re now looking at is then: less than -.033247 and greater than .034361. We’re sitting on eight outliers, then, four negative and four positive. (Two of the negatives barely eek outside of the 2 standard deviation limit, but for the standards of completeness, I’m including them.)
Let’s look at the positives first, those whose Actual Outs exceeded Expected Outs. They are: Rheal Cormier (DER Difference .03737), Paul Quantrill (.04219), Felix Heredia (.04822), and Octavio Dotel (.06265). All relievers. An expected out range of 144.2 (Dotel) to 191.8 (Heredia). As far as I can tell, there’s no discernible correlation between Expected Outs and DER Difference. I’m going to chalk it up partly to good defense and partly to pure chance. (I know I shouldn’t do that, but I can’t think of a real reason for these guys to have such good defense behind them, specifically).
Now let’s look at the positives, starting with the two marginals. More relievers, in the personages of Antonio Alfonseca (-.03337) and Tim Worrell (-.03454). Worrell’s appearance here surprises me, because he pitched well all season, apparently despite his bad defense. And now for the two larger outliers: Aaron Heilman (-.04166) and Jeff Weaver (-.05102). Usually Weaver’s the posterboy for pitchers getting hurt by his defense, in terms of DIPS analysis (although in Baseball Crank’s July analysis, Glendon Rusch showed up as the posterboy, and he’s only sitting at -.02054, good for 28th worst. How much of that is due to second half starts not included in Crank’s data, I don’t know). To emphasize Weaver’s performance (since he’s the only outlier over 300 innings), his Expected Outs – Actual Outs = 29 Outs. The next close on the negative side is Andy Pettitte (-.02294) with 14.8 outs lost, on the other side of the ledger, Jason Schmidt (.02841) gained 15.5 outs from his defense according to this model.
While Weaver’s appearance on the list is interesting, Heilman’s is much more shocking. The conventional wisdom is that he sucks and needs to go back to AAA. However, he did lose 8.6 outs (137 expected) to his defense (I’d probably blame, in order: Roger Cedeno, Robbie Alomar, and Joe McEwing. The Mets do keep showing up near the bottom of David’s studies, if you look at some of the other data sets). He may have just suffered a string of bad defense.
Next, I’m probably going to take my data set and arrange it by team, to see if there are any trends there.
Bye for now.
JH 9:17 PM