Sounder at Heart is, in general, a statistics/analytics friendly blog. Two of our writers focus on the statistical aspects of the game and most of the rest will apply the work of others while not developing our own metrics (anymore). It is easy to get lost in the available data. Quality judgments are difficult, but the limited data do seem to be able to describe strategic and tactical methods at the club and league level. One of the stories in aolsh's plan is to use the data to show club differences within the league as well as finding twins between MLS and Euro clubs.
But I'm not a statistician. I tend to apply others work, or re-present it in a way specific to the Sounders and/or MLS. There is a danger in this. One can dive too deep into the data and either lose their way or discover oddities that do not exist.
Seattle Sounders FC has an analytics expert. Ravi Ramineni works on the training team with a primary focus on sports science to improve match readiness and fitness. Last week I chatted with him about a few things related to data and soccer.
SaH: Is data currently better describing style or quality?
Ravi: There are a lot of nuances. Part of it has to do with what style of play the team has. At the basic level, when you are measuring something you need to know what you are trying to do. For example you take a team like Barcelona. They go about passing the ball around creating mismatches and take the most opportune shot. It doesn't work for other teams. Take Levante which is a just a counter-attack and pump the ball to Oba and get a goal. You have to look at those and then look at the team's style and then look at what stats make sense for it.
Just looking at match stats at the end of a game, at the game level, you can come up with conclusions and most of the time you might be right, but it's purely chance unless you understand exactly what we are trying to do because it becomes if we are trying to get the ball early to our forwards. Did we do that enough? Is there even a stat for that? Sometimes you can measure it and sometimes you can't. I would say it is very contextual.
SaH: Is the big data for soccer overwhelming analysts?
Ravi: It's not big data yet. We're talking about kilobytes, maybe megabytes per game. Even if you look at GPS data it is probably megabytes per game. I would say that is not big data. I worked in big data that was terabytes of data. If you are trying to extract useful information about it you write a program that you need to run for a day, or eight, nine, ten hours and you get smaller quantities so you can look at it in excel or something.
If you add up all the data, the GPS, the video (that's big), if you add it up then you're talking big data, but the stats you see they aren't big data.
This chat is a small reminder that the data can tell the story, but only if you know the story you are trying to tell. It's kind of like verbs and nouns. It is neither a problem, nor a solution on its own.