Sports Economics

Ever since I read Moneyball by Michael Lewis, I’ve been tangentially interested in the economics of playing sports. In a nutshell, Moneyball describes how a baseball team (the Oakland Athletics) used statistics to better predict how successful any given player would be, rather than traditional methods such as “he looks good in a uniform”. This new approach helped Oakland be more successful than they had been previously, and also helped them continue to be successful when larger teams such as the New York Yankees cherry-picked their best players by offering ridiculous salaries.

It turns out that this sort of economic analysis has been going on for over 30 years in baseball, and is known as sabermetrics (see Baseball Prospectus for some examples). Similarly, American football is analysed on the same basis (check out Football Outsiders).

A couple of days back, Malcolm Gladwell posted about some analysis being done on basketball by Dave Berri. Berri believes that performing these analyses can help understand how to model human behaviour throughout economics.

After I read Moneyball, I mused on a football forum about it and how it could be applied to football, but was shot down both by people who didn’t believe you could analyse football in such a way (just as people shot down the baseball and American football analysts) and by people who didn’t believe that Moneyball and the success of Oakland was all it was cracked up to be (despite the fact that many other baseball teams now use similar analyses).

I don’t know if anyone else has been similarly inspired to do this or not, but I think it would be interesting to see. Originally, I thought that the biggest problem was getting hold of enough data to perform any kind of meaningful analysis, but I think a good place to start would be with the minute-by-minute match feeds that the BBC use:

I think these actually come from AP, or a similar agency, but it seems to me that you could begin to build up a reasonable database from these feeds. It would then allow you to start measuring defensive and offensive efficiency. Of course, the feeds are missing large amounts of data (which may or may not be available from Opta) that would be useful, but I think using them to start with would be a great starting point.