The party line on soccer analytics goes as follows (and I’ve often pedaled it myself): soccer is a complex team invasion sport in which ‘discrete events’ involve a complex array of mutually-dependent variables, most of which require a lot of resources to record and even more to disassemble and analyse into applicable metrics. Because of this, companies and clubs who gather/analyse this data are loathe to publish their findings.

Or not.

I’ve become a big fan in recent months of James Grayson’s analytics blog, one of several publishing findings based on data given to them by major firms like Opta to play around with. While media pundits with large audiences and enormous Twitter follower counts preach about how “Moneyball won’t work in soccer,” Grayson quietly alerts his 132 Twitter followers to those soccer statistics with reliable, long-term predictive value, that don’t fall prey to regression to the mean.

A post from last week offers one of the better examples. Here, presented simply and clearly, is a bona fide metric. Yes, one that will likely have to be revised based on improved data gathering on shot type, but a predictive metric nonetheless: Total Shots Ratio, or TSR, which is formulated as (Total shots for/(Total shots for + Total shots against)).

The stat generally reflects a highly generalized skill: “ball control.” As Grayson writes, “The take-away message is that teams that control the ball will gain better results and thus more points.” This may not seem like a groundbreaking revelation, but it at least confirms that reactive tactics don’t generally succeed over the long-term. Hence, the 2011-12 Chelsea probably doesn’t have legs going into next season. Or, perhaps more importantly, far better to engineer a Swansea-like approach within your promoted club than play for ugly draws.

In fact, Grayson applies his methodology to Stoke City, a team that has made its name on Premier League survival by attrition, and has them finishing just above the relegation zone.

Perhaps the relative success of the promoted teams last season came from an advanced understanding of the long-term ineffectiveness of reactive football. Just a thought. In any case, the work of Grayson and others like him is a reminder that analytics need not be the shadowy work of club performance analysts or Infostrada number crunchers.

Comments (2)

  1. One simple question about analytics: How many matches’ worth of data do you need in order to have a sample size large enough to make any determinations?

    SB

Leave a Reply

Your email address will not be published. Required fields are marked *