The Baseball Analyst Issue 4: a review

A scan of this newsletter is available on the SABR website, as are the other 39 in the set.

Bill James gained a bunch of new subscribers with this issue (one of whom was me), so he used the introduction remind his readers of the newsletter’s purpose. Basically, he was looking to develop a sabermetric community, and saw the Analyst as a vehicle to give that community a voice by offering to publish interesting work. He also offered some guidelines for prospective writers, and promised the readers that the content would be better than the newsletter’s production values.

John Billheimer looks at ways to compare starting pitchers with relievers. His preferred method is Pro-rated ERA (PERA), which divides credit for inherited baserunners between whatever pitchers have a hand in the runners’ advancement (or lack thereof). This obviously-sensible adjustment still makes the best relievers look superior to the best starters, but that’s a separate issue and not discussed in any detail. Billheimer also plays with something he calls SLIPs, which are essentially Blown Saves; he dismisses this stat as not helpful. Odd, really, since nowadays we track Blown Saves but do nothing to better apportion pitcher runs. An excellent essay, all in all, with his methods explained very well.

Dallas Adams returns with an update to his Issue 1 study of run distributions by adding data for the (higher-scoring) 1977-81 seasons. He then begins to examine ways to extend his study to the inning and at-bat level. This is a statistics-heavy study, and is completed in a subsequent issue. What’s there is interesting, but heavy reading.

Mark Lazarus takes a look at the defensive support received by major league pitchers, as measured by error rates in 1982. He’s aware of, and discusses, the weaknesses in this analytical method and his one-season data set. Nonetheless, this study turned out to be far more interesting than I expected. The anomalies reported in the data are especially interesting. This topic deserves more study. Not sure that I’ve seen such a work.

Finally, Jim Reuter improves Isolated Power, which he adjusts by removing hits from the denominator (a useful, but not trivial, change; the data’s available but not an in-the-head calculation). The results are interesting, though generally predictable. My overall impression of Reuter’s work remains that he really needed to give his calculations names which didn’t duplicate common usage.

This is the Analyst’s best issue so far. It will be interesting to see where Dallas Adams’ piece takes us in the next issue.