The analytics of soccer
Posted June 10, 2010
Imagine this, soccer fans. It's a zero-zero tie in the 89th minute of the England-US match of the World Cup match. A US defender is called for a (very controversial) penalty. English striker Wayne Rooney steps up for the penalty kick. He faces goalkeeper Tim Howard.
Now, statistically speaking, which way does Rooney kick most of his shots? Lower left, upper right? What's his success percentage on each one? Which way does Howard most often dive? Analysing these stats, should Rooney and Howard be able to optimize their chances?
While sports like baseball and football have been measured and optimized nearly to death, soccer is still very much a work in progress. I talked about this yesterday with Bruno Aziza, an exec at Microsoft Business Intelligence, author of Drive Business Performance, and a die-hard French soccer fan. He's accumulating all sorts of stats and analytics about soccer in preparation for the World Cup.
Soccer is much tougher to measure, of course, because it's more of a flowing game, and not, like baseball and football, a series of plays or pitches. (The penalty kicks are the exception, and they lend themselves to analytics.) Basketball is a flowing sport, much like soccer (when the refs aren't calling gazillions of fouls). But the higher number of points provides more opportunities for metrics. Still, soccer analytics are on the rise. One software company, ProZone, sells pro teams yearly subscriptions to soccer stats for more than $200,000. Here's a thoughtful blog post on it.
In the podcast, included here, Aziza argues that soccer is more like business. It's harder to figure out exactly what to count. But it doesn't mean it's not worth the effort. Insights can surface once you start trying. In fact, for a team like the U.S., with a roster of lower-voltage stars, analytics might make all the difference.
Link to original post
Other Posts by Stephen Baker
The moderated business community for business intelligence, predictive analytics, and data professionals.