Thursday, February 26, 2009

Use same criteria for batters and pitchers.

A friend of mine has long contended that Pete Alexander was better than Christy Mathewson.  Recently he expanded that to Alexander was better than Walter Johnson.  I showed him that ERA+ (earned run average adjusted for era and park) showed that Alexander and Mathewson were maybe even but that Johnson was clearly better than either.  He countered that ERA+ is not everything and that we should not rely on only one stat.  I agreed but I also knew that he was dodging the issue.  OK, he's a lawyer, so what can you expect?

I challenged him to attack ERA+ but have not heard back on this.  He never admits mistakes, like his long standing and discredited contention that  Joe Torre was the reason that the Yankees have not won the World Series since 2001.  Torre in his first season with the Dodgers won the first round of the 2008 playoffs, something he had not done with the Yankees since 2004.  The lawyer thing again.

I am not nuts about ERA in general.  Two outs, batter reaches on an error and any subsequent runs are unearned.  That does not accurately represent the pitcher's performance.  Some pitchers seem to wallow in it.

There is general comfort in using OPS+ to judge batters.  It is on base average plus slugging average adjusted for era and park.

Why not use OPS+ to evaluate the performance of pitchers, too?  Not because it's clearly better than ERA+ but because it's at least as good and because it would be the same for both batters and pitchers.

Now that people contributing to retrosheet.org are providing the missing values (doubles and triples allowed) for this equation we can determine the results.  There would be much more symmetry in comparing performances of batters and pitchers.  I'd like to know the percentage difference between a player's OPS+ and the league average.  I'd like to know if great pitchers have a higher percentage difference than great batters.

Let's use the same criteria to judge batters and pitchers.