There are far better sports to apply advanced metrics to than soccer. Basketball, for example, features loads of statistical categories and tons of points. In American football, everyone lines up in position after each down, making it relatively simple to run simulations. And baseball is so stats-friendly that it's often more fun to argue about changes in WAR and hat sizes than to watch the actual games themselves. The beautiful game, on the other hand, features frenetic action but relatively little in the way of discrete, quantifiable occurrences. Even soccer's results are maddeningly obtuse; the most common score is a 1-1 tie. Slowly, though, the world's most watched sport, which features most of the world's most popular athletes, is giving up its secrets to the quants.
"In terms of possession and touching the ball, soccer players are closer to offensive linemen than to quarterbacks or running backs."—David Sally
Leading the way are David Sally and Chris Anderson, college professors turned sports consultants who just published the book The Numbers Game: Why Everything You Know About Soccer Is Wrong. In it, they argue that much of the conventional wisdom about the value of strategies and players doesn't hold up to rigorous analysis. (The title of the Dutch version of their book, for example, can be roughly translated to "You should take short corners.")
In the following interview, edited for clarity, Sally and Anderson talk to Gelf about the importance of managers, the potential backlash to their work from stats-resistant soccerphiles, and bringing their geekery to the masses.Gelf Magazine: Has there been any backlash to your research in the soccer community like there was to sabermetrics in the baseball community? Is there a fútbol version of Joe Morgan running around saying that you guys can never capture the importance of Fernando Torres's clutchness?
David Sally: It's still too early for there to be "backlash," which I think of happening when the old order is really threatened by a fully emergent new movement. Parents battle and ground their teenagers and not their toddlers (10 minutes on the step will do!). Analytics in soccer isn't far enough along to truly threaten anyone yet. There have, however, been a lot of skepticism and a little bit of gloating when certain football men bring up the struggles of the Fenway Sports Group with Liverpool and the failure of Damien Comolli to reproduce Theo Epstein's success. There are multiple potential "Joe Morgans" out there, like the manager of a club in the Premier League who, as we detail in the book, responded to our analysis with the retort, "You can't measure the size of a man's heart." Actually, as we like to point out, you really can: x grams of muscle mass pumping y ccs of blood volume per minute!
Chris Anderson: I agree; I'd only add that it's hard for there to be a backlash when the metrics, analytic insights, stats, etc., are still relatively undeveloped. The soccermetricians out there are less numerous and less far along than the sabermetric community was by the time baseball teams started to take their ideas seriously. It's also hard for there to be a backlash when it's a game that's more complex than baseballthis means that, for practical purposes, it's easier to explain away outcomes with anecdotes or intuitions or one-offs.Gelf Magazine: What aspects of your research are you worried have gotten dumbed down or misconstrued as your book has been reviewed in the popular press?
David Sally: I have to say, not much. Either we wrote very very clearly (unlikely) or the reviewers and readers have been well primed by analytics in other sports and by similar books and articles (e.g., Jonah Keri; Joe Posnanski; Simon Kuper; Jonathan Wilson; et al.) to be sophisticated consumers of evidence, social science, and data-driven conclusions. No one has put words in our mouths or made us out to be extremists who advocate always taking short corners, who state that the game is all luck and that superstars don't matter.
Chris Anderson: Well, a few people think that we said Chelsea literally should have bought Darren Bent. Our analysis of his goals was meant to show that he was extremely good value for the money, and he had a knack for scoring the goals that won games. That, presumably, should be part of thorough analysis when you're buying a striker, but that's not how strikers are evaluated by teams or fans. Of course, buying someone is about more than just which goals he scores.
And a few people have been upset by the subtitle, understanding it as us telling them they don't know the game they love. What we really meant by that was that we're all wrong about some things, and the best way to find out is to ask lots of questions and look at the evidence.
David Sally: That's a tough one especially because we can't yet quantify his full production on the field either. Factors such as leadership, role modeling, and effort raising are among those things that count but can't (yet) be counted. The point is for coaches and managers to know about these factors and use proxies or estimates to make smarter tradeoffs when assessing a lazy or extra-egotistic or "toothsome" striker.
Chris Anderson: Sure, you can count how many shirts he sells in the club store, how many fans come to games because of him, etc. In terms of the "intangibles"that's the trouble with intangibles: I agree with Dave that they're harder to get your hands on and head around. But you can approximate a lot of things with tools developed by psychologists. It's a job performed as part of a teamthere are lots of those out there in the real economyand evidence for individual "performance" as part of a team can be found. IBM, Google, Pfizer, and so on do it every day.
Gelf Magazine: You state that managers don't affect much. Are national teams different in that regard because the manager has much more control over the lineup, etc? What should we make of what Jürgen Klinsmann has done with the US Men's National Team?
David Sally: To play a Clintonian game, it depends on what the meaning of "much" is! Managers are less influential than the history and infrastructure of their clubs and than the size of the overall payroll. But, they may be, in some instances, as important as their striker or their entire back line. Through skill improvement, motivation, tactics, and so on, the manager has a significant impact on the performance of the team. Roger Bennett quotes Trapattoni as saying, "A good manager makes a team 10% better and a bad manager makes it 30% worse." That fits roughly with the evidence we review in The Numbers Game.
As a rule, managers aren't more influential on national teams than they are in their clubs. We write a lot about English football where the vast majority of teams are run by managers who both coach and recruit talent. The Premier League is a league full of Mike Holmgrens and Bill Parcells, managers who are also GMs. Even in the other European leagues where clubs have sporting directors who fill the general manager role, the manager still has an outsized voice (equivalent to Bill Belichick's) in personnel matters.
Chris Anderson: On a practical level, being the coach of a national team is also a somewhat different job. They see their players much less often and so get much less of a chance to work with them as a team. They select and coach players who aren't normally teammates. And the games are interspersed throughout the year. Finally, the most important competition (the World Cup) only comes around every 4 years. To me, the great national team coaches are those who are able to forge a real team spirit during the highly charged few weeks of the World Cup, which resembles March Madness much more than normal league play.
Gelf Magazine: Which of your findings was most surprising to you, or at least went against your experience as players and viewers?
David Sally: For me it was the fact that a player on average has contact with the ball for only 53 seconds of a match. In terms of possession and touching the ball, soccer players are closer to offensive linemen (non-centers) than to quarterbacks or running backs.
Chris Anderson: There are lots of stereotypes about soccer being different in England compared to, say, Spain. I was surprised by how similar soccer "production" looked across the top leagues. If we didn't know which league we were analyzing, you couldn't tell them apart just by looking at the numbers.
Gelf Magazine: What aspects of the game do we still have the most to learn about from a statistical perspective?
David Sally: In keeping with the previous answer, the true game all happens in space away from and without the ball. Measures of effective runs or defensive shiftings or penetrating triangles are nowhere to be found. It is an analytical problem shared by all the flowing team sports (not baseball, not cricket), but is of utmost importance in soccer.
Chris Anderson: And I'll give the more traditional answer: everything. There is lots we don't know even looking at the more basic game data and ball events. As we say in the book, if it were medicine, we'd still be in the age of leeches.