Moneyball Cubed

Today was a good day for Bay Area baseball fans. The Giants salvaged a game in their series against the Pittsburgh Pirates, and more dramatically, the A’s swept the Yankees in Oakland, winning the final game of the series in the 18th inning on a bases-loaded single by rookie Nate Freiman against the all-time great reliever Mariano Rivera. It is satisfying to see the team with the largest payroll in baseball, the Yankees, losing out to the team with the fifth lowest payroll, the Athletics. How to do well with few resources is of course the theme of the book on the A’s from a decade ago, Moneyball, and it seems the A’s, or maybe in particular their general manager Billy Beane, have found some tricks to doing this again.

Last week I had a glimpse into what baseball analytics is like today in the era of big data. I attended a portion of the Semantic Technology and Business conference held in San Francisco and learned about one company’s work on applying semantic big data technologies to baseball. Vince Gennano, who is president of the Society for American Baseball Research (a.k.a. SABR, after which the term sabermetrics gets its name) gives further details of this work in his writeup here describing how big data allows for the creation of cluster diagrams (see the illustrations in the prior link) mapping out the similarities between different pitchers. Gennano says that when Moneyball was written, there was 2% of the data on baseball player performance there is now. To a good extent this is due to the use of technologies such as PITCHf/x, HITf/x, and FIELDf/x that are tracking the flight of every pitch, every batted ball, and the movement of every fielder in exquisite detail. With that much data being generated, it takes serious analytical know-how to make use of it all, and teams that build that capacity will gain an advantage. Not surprisingly, the company at this conference would not name which major league team (or teams?) they are working for. An arms race is afoot.