Commentary

Why the Angels are overperforming

The Los Angeles Angels are once again beating their Pythagorean win percentage

Originally Published: August 28, 2011
By Carson Cistulli | FanGraphs
Mike Scioscia Jim Cowsert/US PresswireMike Scioscia 's Angels are right back in the middle of a pennant race again.

In the 1980 edition of his Baseball Abstract, sabermetrician Bill James presented an idea that was as intuitive as it was surprising -- namely, that it was possible, by looking merely at a team's runs scored (RS) and runs allowed (RA), to predict (within a few games, at least) what a team's record would be. James referred to the result as a team's Pythagorean win-loss record, owing to how the formula [RS^2 / (RS^2 + RA^2)] resembled the Pythagorean theorem to which most every student is introduced in middle school math class.

Perhaps even more extraordinary than merely estimating a team's present record, James' work also showed that a team's current Pythagorean win percentage is better than a team's actual winning percentage at predicting that same team's future winning percentage. Run production and prevention, James taught us, are at the heart of a team's wins and losses.

The strength of Pythagorean win percentage is that it strips out luck and randomness. Consider, for example, two totally fake teams -- we'll call them the Nerds and the Geeks -- playing a weekend series against each other. Let's say that the Nerds win the first two games, each by a score of 2-1. Now pretend that the Geeks win the third game, 25-0. Despite the fact that the Geeks have outscored the Nerds 27-4, their record is only 1-2. Now it's possible that something weird has happened -- that, for example, the Nerds got locked in their mothers' basements and had to send out a replacement squad -- but it's more likely that the Geeks are the better team, but just happened to have bad luck. Their Pythagorean win percentage would reflect this.

Our understanding of Pythagorean records has only improved since James' original discovery. Further research shows, given the run environment of recent years, that 1.83, and not two, is the preferable exponent. Other research shows that the majority of teams will finish within four games of their Pythagorean win total.

Curiously, for all of the increased understanding of win estimation, there's a mystery that has baffled baseball nerds everywhere -- namely, the ability of the Los Angeles Angels to consistently win more games than their Pythagorean record would suggest.

Consider, as Exhibit A, the Angels' win (W) and Pythagorean win (PythW) totals since 2004 (and through Friday), with the differences between the two over that time.