Interleague play is over for 2009. To the surprise of no one who follows baseball the American League dominated the National League yet again. This is part (but not all) of the basis for the argument that the AL is the stronger league than the NL. This angers some NL fans, who think it is baseless, and who sometimes point to isolated interleague series where an NL team beats the AL, or get angry at the results of Justin Inaz's league-adjusted power rankings at Beyond the Box Score. I can't cover all this stuff in one stroke (nor would I know how), but I thought I'd make a short post briefly getting some perspective on the results of interleague play the last few years. Let's start with a nice, nerdy graph.

[Click on the picture for a larger view]

The graph is worth a thousand words... The green line is pretty obvious -- it gives the winning percfentage of the American League in interleague play from 2005-2009. The red line is the "expected" winning percentage. This is derived from a variation of Bill James' Pythagorean Theorem, which estimates how many games one can expect a team to win give the number of runs the team scores and allows: Runs Scored^2/(Runs Scored^2+Runs Allowed^2).

The variation I used above is PythagenPat, which gains in precision by depending on the run environment a different exponent (the "2" in the basic version given above) depending. That exponent is derived by runs per game^0.287.

For each season's series of interleague games, I took the total number of runs scored and allowed (I'm sure someone has done this before, but I couldn't find it after a few google searches, and even then I could only locate the RS/RA numbers back to 2005, otherwise I would have done the whole thing), then divided it by number of games played to get the runs per game, then get the exponent, then derive the winning percentage.

The table below lays out the data. The four columns on the left given the actual interleague records each season from 2005-2009, then the winning percentage, then, for perspective what that winning percentage would mean over 162 games. The four columns on the right give the runs scored by each league, and again, the AL's expected winning percentage based on PythagenPat for that season, and then that winning percentage over a 162 game season.

The results speak for themselves -- not only has the AL dominated the won-loss record each season, but according to PythagenPat, in only one season (2006) did they "outplay" the run differential, as one can see in the graph given above. The final row treats all five seasons as one -- the AL has had a .566 winning percentage over the past five season in interleague, and an expected winning percentage of .576. Again, for perspective, the AL "team" over the past five years has been a 92-93 win team in interleague, which means the NL has been around a 69-70 win "team." I'm not an expert on statistics in general, but I think 1259 games is a pretty substantial sample size. If there where just two teams in baseball, and they played each other 162 games a season, and for five years in a row, one team won around 92 games, and the other won 70 games, would you be comfortable saying that the 92 win team was clearly superior?

As I mentioned above, this is not the entire basis for separate WAR adjustments for the leagues or for adjustments to different kinds of "power rankings." Nor does it mean that players like Albert Pujols and Tim Lincecum are "no good." We can still be fairly sure than Lincecum is one of the best pitchers in baseball, and that Pujols is probably the best overall player. But I do think it's makes a pretty strong case for the current superiority of the American League.