Here’s the first part of a brief collection of observations from this weekend’s MIT Sloan Sports Analytics Conference, held at Hynes Convention Center in downtown Boston:
• The conference was structured around a series of four- or five-person panels, many of which dealt generally with the place of analytics within one of the major sports. From a statistical standpoint, the hockey analytics panel was pretty underwhelming, featuring mostly “establishment” professionals beating up on the lone analytics advocate. Toronto Maple Leafs GM Brian Burke, while infinitely charming (“Statistics are like a lamp post to a drunk: useful for support, but not for illumination”), had little patience for the ideas of Michael Schuckers, an associate professor of statistics at St. Lawrence University. NESN analyst Mike Milbury joined him in the assault, asserting that the “Moneyball” method holds limited value because it has never yielded a championship, pointedly ignoring the two captured by the Red Sox and Theo Epstein, decided stathead.
It’s a shame, because, of the four major sports, hockey receives the least attention from the analytics community. Whereas research in baseball, the sport best understood from a statistical perspective, has already uncovered many ground-shaking discoveries and forced people to rethink the sport, hockey represents an opportunity to return to the Bill James era of innovation and essentially learn the game again. Occasionally, a tidbit of interesting information emerged; for instance, Schuckers’ research indicated that a faceoff win is worth the equivalent of only one one-hundredth of a goal, and he stated that teams can control the number of shots they face, but typically not the quality of those shots. However, he never got the chance to fully make his case, given his unwillingness to engage the other panel members in challenging traditional hockey notions.
• Beyond the obvious questions, like when football coaches should go for it on fourth down, the coaching analytics panel presented a couple of interesting ideas. During a discussion in which panelist Bill Simmons bemoaned the tendency of teams to “recycle” old coaches -- rehiring previously fired coaches rather than giving an up-and-comer his first big break -- former coach Jeff Van Gundy argued that the general manager should be held to the same standards of performance as the head coach. Van Gundy speculated that, on average, a GM gets to hire (and subsequently fire) an average of two or three coaches before their job becomes endangered. It’s a fair point; the coach can only work with the players he’s given.
• The basketball panel highlighted one of the most daunting challenges for analytics to maximize its effect on sports: communicating analytical ideas to those most directly involved in the games, the coaches and players. Obviously, players (and, by extension, the coaches) don’t feel like their performance is bound by some predictive pattern, so it’s difficult for them to buy in when, for instance, they’re told they dribble too often before they shoot.
As with all other problems in life, there’s an app for that. Multiple executives and coaches across sports referenced the iPad as a simple method for illustrating concepts through film study. Show a player all the times in sequence he overdribbles before missing a shot, and he’s much more likely to be convinced. The evidence is staring him in the face, in an easily digestible format. Of course, as a coach, one doesn’t always have to resort to the most honest methods to get through to one’s players; Van Gundy also said he wouldn’t be averse to fabricating statistics entirely if it helped him get a useful point across to one of his players.
There have been other success stories, too. The panel briefly discussed the role of Mavericks assistant coach Roland Beech, whose previous affiliation with professional basketball was limited to his role as founder of 82games.com, a website devoted to statistical analysis of the NBA. The Mavericks, along with the Rockets, have been credited as pioneer franchises in implementing statistical analysis into their cultures, and the Mavs’ hiring of Beech has certainly paid dividends. Analysis of lineup combinations and matchups was largely responsible for the decision to start J.J. Barea against the Heat in the NBA Finals, a move crucial to Dallas’ victory in six games.
• At the end of the conference’s activities on Friday, Bill Simmons conducted an excellent live podcast with baseball statistics godfather Bill James, whose work analyzing baseball predated the current statistics craze by three decades. The amount of monotonous work that must have been involved in clipping out and hand-tallying countless box scores from newspapers is simply staggering. His ideas went unrecognized for years, yet he persevered despite very limited circulation of his annual Baseball Abstract and a day job as a factory worker. I’m not sure exactly what duties James currently performs in his role as advisor for the Red Sox -- perhaps just being Bill James -- but it likely involves the secret to making a name for oneself, which he revealed to the many aspiring general managers in attendance in one pithy quote: “You have to know something not everyone else knows.”
The author is solely responsible for the content.
Stats Driven features a closer look at statistical analysis, sports strategy and trends within Boston sports. Andrew Mooney, a student at Harvard College and an active member of the Harvard College Sports Analysis Collective, is the primary contributor. Email him at firstname.lastname@example.org and follow him on Twitter at @mooneyar.