Officially, Saturday was the 2009 MIT Sloan Sports Analytics Conference. Unofficially, it served as the third annual APBRmetrics convention. The day was the latest example of the growth of the NBA's statistical analysis community and its newfound influence on the league. Taking note, reporters from The New York Times, CBSSports.com, SI.com and event sponsor ESPN were in attendance to chronicle the proceedings.
Certainly, other sports had a presence over the course of 12 panels and three presentations. I enjoyed the Baseball Analytics panel that featured Baseball Prospectus' Christina Kahrl, as well as a session on Talent Identification that featured Aaron Schatz of Football Outsiders and the Pro Football Prospectus annuals (and even Michael Forde from the Chelsea soccer club). Still, make no mistake--this was an NBA conference first and foremost. By my unofficial count, 12 different NBA franchises were represented at MIT's Stata Center to go along with nearly all the prominent statistical analysts who are not associated with teams. Included amongst that group were a pair of general managers (co-chair Daryl Morey and Portland's Kevin Pritchard), an owner (Dallas' Mark Cuban) and no fewer than eight people in the employ of Morey's Houston Rockets (quoting ESPN.com's John Hollinger: "Fortunately for Daryl, there's no luxury tax for analysts.")
From that perspective, the day's highlight was the Basketball Analytics panel moderated by ESPN.com's Marc Stein and made up of a reluctant Morey, Cuban, Hollinger, Dean Oliver of the Denver Nuggets and Mike Zarren of the Boston Celtics. Because the teams who make extensive use of statistical analysis believe it to be a competitive advantage, none of the employees wanted to be especially forthcoming. Morey was particularly circumspect.
Fortunately, Cuban was there to liven things up. While he did not share the specifics of how the Mavericks use numbers, Cuban did offer additional insight into the Jason Kidd/Devin Harris deal in the midst of an impromptu debate on the trade. Cuban was willing to essentially admit that Harris has the brighter future while arguing that Kidd better fit Dallas' timeline and implying the value of the trade lay in having Kidd for two years and then taking advantage of his expiring contract.
Cuban also sparred with Zarren over which of the teams had the better offer for Kevin Garnett and offered his usual critique of the NBA's refereeing, arguing that three of the 13 people on the court had 80 percent of the impact. When Cuban, discussing teams' reluctance to share any of their analytical work with competitors, said he definitely doesn't give anything to his division rivals in Houston, Morey shot back, "Except referee ratings."
As anticipated as the Basketball Analytics panel was, the surprise star of the show ended up being the lunchtime presentation exploring the phenomenon of the hot hand by John Huizinga of the University of Chicago and Sandy Weil. The hot hand is one area where the numbers have strongly disagreed with conventional wisdom. The definitive study on the subject was performed by academics Thomas Gilovich, Amos Tversky and Robert Vallone, who analyzed the shooting trends of players from the early-'80s Philadelphia 76ers, the free-throw performance of the contemporary Celtics and also performed studies in a more controlled environment using the Cornell basketball team. None of these studies showed any evidence of a hot hand. None of that stopped the NBA Jam series from adding the "He's on Fire" feature, and we often hear of players who feel they've entered "the zone," a state in which they simply cannot miss.
Using play-by-play data collected by Weil, Huizinga had the insightful thought to explore the issue in more detail, looking at how players respond to their own perception of being "hot" after making a shot. They found a decrease in player shooting percentage after a make, much of which can be traced to an increased tendency to shoot jumpers instead of taking the ball to the basket. To answer whether this could be related to defenses taking note of the "hot" player, they then studied how quickly players attempted another shot based on whether they were coming off a make or a miss and found that hitting jumpers in particular led them to become quicker on the trigger--especially for guards who have the ball in their hands and can more easily call their own number.
There is a certain conventional-wisdom justification for these findings. Color analysts frequently note that the best thing for a defense is for a weak outside shooter to make his first shot and fall in love with his questionable jumper. In this case, many of the players--limited to go-to scorers to provide adequate sample sizes--are in fact fine shooters, but their apparent willingness to take lower-percentage shots after a make still benefited the defense.
A question left unresolved by the study is how much of a role coaching plays in the findings. Are teams running more plays for their stars after a make? One way to look at this issue is to look at a group of lower-usage players unlikely to get plays, potentially worthwhile despite the sample-size concerns.
You can check out the presentation for yourself at Weil's Web site.
Besides the hot-hand study, the other key observation I took from the conference was that an interesting frontier for future exploration for analysts in all sports is taking their work to the player level. That is, will players eventually be able to use numbers for the benefit of their own games? We already saw some evidence of this phenomenon in Michael Lewis' Shane Battier profile, but it was made clear that Battier is the only Rockets player who uses the analysis generated by the team.
During the Baseball Analytics panel, former Houston Astros GM Tim Purpura mentioned that he would use information about pitchers' success on various counts to convince them of the importance of getting ahead. Meanwhile, Cuban used the example of players' shooting percentage from different spots on the floor, saying he'd almost always see the player slide over to where they were at their best not long after a conversation.
Dealing with players is an extreme example of what Zarren pointed out, that communication skills are often more important for analysts working in a team environment than training in formal statistics because of the need to explain findings to the busy decision-makers on the coaching staff and in the front office.
Of course, much of the value of the conference came outside the classrooms and in the hallway, where members of the APBRmetrics community got a chance to chat casually--a process that continued afterwards in a reception and then on into the night at various Boston watering holes. The SABR Convention has long served as an annual opportunity for baseball analysts to reconnect, and the Sloan Conference has filled the same need in basketball. For friends and acquaintances who have interacted online, it's a valuable chance to put a face to a name.
In the span of three years, the Sloan Conference has quickly become a can't-miss event for the NBA's analytical community. That doesn't figure to change any time soon, especially if Lewis ends up headlining next year's conference, as Morey indicated in a podcast Monday with ESPN.com's Bill Simmons (yet another famous panelist). I know I'll be there.
Kevin Pelton is an author of Basketball Prospectus.
You can contact Kevin by clicking here or click here to see Kevin's other articles.