Prozone and its use in soccer analysis
Categories: Match Data Collection, Team Performance
A few weeks ago I added a link on the blogroll for Prozone, which is a company that produces a software package that does performance analysis for soccer and other sports. I remember reading something about the company when it was founded, but I didn't remember their initial software being so sophisticated. At any rate, their third version of ProZone and current version of MatchTracker appear to be able to produce the type of detailed statistics from video analysis that I originally thought was some ways off. Here is an example of Prozone's match analysis, in this case a review of the USA vs. Mexico World Cup qualifier (H/T Soccer by Ives). I wouldn't be surprised if a more detailed analysis is being kept under wraps by the national team staff.
So we can generate more detailed data than we were able to in the past. Now the next task is to determine whether we can develop better statistics that take into account these data. It would also be nice to have an open-source clone of Prozone — that software has to be hideously expensive!
UPDATE: Sarah Rudd mentions in the comments that there is a competing package called Amisco (warning: the website takes a very long time to load works fine now). I hadn't heard of it, but I read abstracts of research papers that have used Amisco to generate data and it looks like it does a good job of generating high-quality time-dependent match data. So as Sarah said, it is possible to obtain very rich datasets from video analysis of football matches.
This website from the University of Calgary has several links to companies that have developed computer-based analysis systems for a variety of sports. There are a few devoted to soccer, but I've only looked at one off that list and I wasn't impressed with the quality.