Apparently, there's some kind of football game going on here in the US this weekend. Strangely though, the ball isn't round. The playing field isn't even oval. No, this is American Football.
Football (of all stripes) and statistical analysis have a long history together, but in each case the focus appears to be different. For UK football (soccer), much of the research appears to be around predicting the outcome of games (usually, in order to optimize betting strategies -- in fact, I have a friend in the UK who's been quite successful modelling games and betting on the outcomes). I don't know much about statistics and Australian Rules football, but a quick google search reveals a lot of work in identifying the causes of injuries. For American Football, though, I've found a couple of articles where statistical analysis is used to determine the optimal strategy for play.
Around the time of the 2008 Superbowl, Technology Review published a detailed and interesting article describing how the New England Patriots use the analysis of the results of different plays to select strategies. Jim Orlin follows up in his blog with more detail about the kinds of analyses employed.
Relatedly, on strategy for building teams, Malcolm Gladwell had a fascinating article in the New Yorker on the wisdom (or otherwise) of recruiting star quarterbacks from college football to play in the professional league. As it turns out, performance in college football isn't a great predictor of merit in the big leagues.
So to the folks in the US and Superbowl fans elsewhere, enjoy the game tomorrow. I'll be watching too, if only for the commercials.
Comments