It’s a little more than a week removed from the MIT Sloan Sports Analytics Conference and, in large part, everything of importance has essentially been summed up by individuals far more capable and qualified.
*note: If you’re interested in more information about the events at the Sloan Conference, I can’t recommend the coverage by the Truehoop Network highly enough.
One of the prevailing themes from the weekend seems to the need to be able to effectively communicate the data that is being harvested from the SportVU system. SportVU is a multi-camera tracking system currently installed in 15 NBA arenas. It can continuously track players and the ball on an XYZ axis at an absurd rate and the consensus for the past couple of years is that the database it builds will be the future of basketball analytics. And so with dense research papers and presentations trying to mold knowledge out of the data, the one thing that struck me was the lack of traditional basketball minds associated with any of the work being done. Not to say there weren’t any basketball personnel in attendance – every team but one was represented by their own staff of quantitative analysts (quants) or front office executives. And several panels prominently featured current and former GMs and coaches.
But the panels had a tendency to deal more in the abstract, the 40,000 feet view of analysis than the details. In the actual research, it was all quants sifting through data, looking for interesting nuggets of information that could either confirm or deny some conventional collected knowledge about the sport. To me, it felt exceedingly difficult to process all the theories that had been derived and be able to process it meaningfully with the way I approach the game. Even Kirk Goldsberry’s presentation – unequivocally the hit of the conference, in part, for its digestibility – seemed to leave me with more questions than answers. (Why are the defenders so good? How is foul rate involved in the defense? How much of the defense was from help rotation versus one-on-one?)
Which isn’t to say that I think analytics is flawed or somehow diminishing our appreciation for the game – not at all. What I left wondering was whether we are approaching the data in the right way. And this goes back to the prevailing sentiment from the weekend – communication is the key. But what if it’s not in the direction we think? The feeling seems to be that somehow the data needs better visualization or storytelling as Goldsberry has done. But I was left thinking:
Do we even know what to be asking to be able to communicate the answer?
Sloan felt almost like an episode of the television game show Jeopardy! In most problem-solving cases, we begin with the question and then go about determining an answer. But in this case, it felt like the reverse. Quants have done such a wonderful job building a database that, like the game show, the answers are only the clue to a question that no one has. And maybe this is where the idea of communication is originating in the wrong direction. Maybe we should be asking coaches, players and front offices about their questions, be able to convert their inquiries into a format that is quantifiable and then refer back to the Jeopardy! board.
It’s wholly possible that progressive teams have already bridged this gap and are effectively mining the data, converting it to a knowledge base able to yield applicable answers to more traditional thinkers. And there’s no reason for them to reveal anything to competitors and give up their edge. But in the public sphere, there are a number of former GMs and coaches similarly trained in the game of basketball that could focus the question in collaboration with quants.
In the Predictive Sports Betting Analytics Panel, Haralabos Voulgaris casually comments in balancing objective data with subjectivity, “..with actually seeing something happen that you’ve seen happen, watching that with your eyes, collecting information, that’s data. It’s just a different type of data, it’s something that you internalize.” It’s the black box nature of the human mind that analytics is trying to unpack, all the hundreds and thousands of processes and correlations we make even before initializing any thought about the things we observe. Basketball experts can take in millions of bits of information and, whether or not they correctly assess the situation, they typically understand what the problem is that requires a solution. This is where it seemed to me where the communication should begin; not at how to convey an answer to coaches and players, but at identifying what it is they are asking.