Raptors coach Dwane Casey is a fan of analytics, and I suspect he would be intrigued with some of the remarkable work being done in determining defensive capability in pro basketball.
The MIT Sloan Sports Analytics Conference wrapped up this weekend in Boston. As the influence of math nerds grows in professional (and amateur) sport, this annual convention has become the go-to event for those interested in the latest advances in analytics.
Toronto Raptors head coach Dwane Casey took over the team’s reins in June of 2011, and has never been shy about expressing how important defense is to him. His players have tended to follow the old business cliche “What’s interesting to my boss is fascinating to me” because they know they won’t get much floor time if they don’t dig in on D.
As the video below elucidates, knowing how to measure NBA defensive prowess is extremely difficult, as there’s precious little data. Nothing drives quants crazy faster than a lack of data on any topic, so the presenters describe their system for accumulating defensive data, and then putting it to work.
You are urged to watch the entire 25 minutes or so, and don’t skip the post-presentation questions from the audience, and the creators’ responses. [20-second timeout: one questioner asks about intentionally “loose defense” but doesn’t use the term “zone defense”; the presenter acknowledges the validity of the question and confesses to difficulty in their model when a player is covering a chunk of the floor rather than a man. I think it’s terrific that the audience was able to spot limitations in the model while agreeing the initial work is sound. Another good question from the audience: how does the model deal with a pump-fake, which led me to wonder how the data would model a shooter attempting a pass when he knows his defender has ruined his chance of releasing an effective attempt?]
So much of defense is context. If Kyle Lowry allows his man to score 23 points, that would qualify as an excellent result against Stephen Curry, but an embarrassment if he’s facing Donald Sloan.
I’d enjoy a Raptors-only examination of the presenters’ created data, as I’m sure I’m not the only one who’d like answers to questions like:
- Is James Johnson truly a sound defender, or does he get extra credit for spectacular events like blocking 3-point shots?
- Is Jonas Valanciunas improving on D, or has he plateaued?
- Why do our competitors run multiple screens to free up Terrence Ross’s man – why is TRoss considered “attackable”?
- How far below an average defender is DeMar DeRozan, and how could he improve?
- Does the presence of a “stretch-4” instead of a traditional power forward negatively impact team D? If so, by how much? If the Raptors gave Bruno Caboclo meaningful minutes late in the season, could we find an answer to my question, or would we not have enough data?
There is no shortage of these types of queries. Although I’m not a mathematician, I appreciate their work. Perhaps next season I’ll get to the Conference and ask a bunch of really smart people some basketball questions.
How about you, Rapture Nation? Are there defensive conundrums you’d like answers to? Let me know in the Comments.