Soccer: On Player Evaluation

•

Voros

I wrote a post on Big Soccer about the challenges that we face in individual player analysis in soccer as compared to baseball.

I guess if I were to suggest a course of action for the statistical community, a soccer version of Bill James “win shares” might be the best idea. This is interesting because I’m not all that enamored of win shares in baseball; it’s interesting but I don’t think any significant improvement on the sorts of things already being done, particularly for modern players.

But for soccer, I think it’s the exact right approach. Figure out the quality of the team the player plays for, and then apportion out credit/blame to individual players based on those results. The difficulty, I think, is that intuitively I believe soccer is a multiplicative sport and not an additive one. That is, if one team has two players rated at 10 and 2 and the other has players rated at 5 and 5, the second team wins 25 to 20 rather than losing 12 to 10.

Granted I can’t prove that, which again is another problem. For the Billy Beanes of the world looking to crossover, they have a very difficult task ahead of them. The good news is that success may prove to bring monumental rewards, because they’d be one of the few people on the planet who would possess that knowledge.

2 responses to “Soccer: On Player Evaluation”

  1. Aaron B. Avatar
    Aaron B.

    I think you’re onto something here.

    I’ve spent some time thinking about player analysis soccer, and I’ll I’ve concluded is that someone would have to do some super-specific scorekeeping/statkeeping (aside from shots taken, goals scored, etc.; things like deflections/poke-aways per game, steals, “good” dead ball kicks, and other obscure stats) for it to have any credence.

    The win shares idea is the best one I’ve heard so far.

    Reply
  2. sidereal Avatar
    sidereal

    The major problem is that the information isn’t there to build ratings on. Sabermetrics is a rich field because there are so many discrete stats maintained that can be diced up and aggregated at will.

    Soccer actually has, I think, all three of the major impediments to baseball-type analysis in other sports.

    1) Lack of stats. The only ones officially maintained in most leagues (to my knowledge) are goals and cards. If you’re lucky you can get into shots and saves. And if really lucky, into something like passing percentage. But even that’s pretty thin gruel to build an evaluation system on.

    2) Players are interdependent. This is a real problem with American football analysis. I can’t rush if I don’t have an offensive line, so my stats suffer, but does that really help evaluate my ability? Similarly, I can’t score as a forward if my midfield can’t get the ball to me, and I can’t get clean sheets as a goalie if my defensive line can’t hold their own.

    3) Results rely on opportunity. In baseball, everyone who gets an AB gets the same chance to do something. In basketball or soccer, I might get a minute on the court or field but if the tactics don’t call for me to get the ball, then I won’t, regardless of my ability.

    I’d love for there to be more rigorous statistical evaluation tools in any sport other than baseball, but I just don’t think the foundation is there, and it will probably be a long time before soccer can even get close.

    Reply

Leave a Reply

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