The slender form of Stephen Curry casts a very long shadow. His Golden State Warriors appear poised to win more than any NBA team ever has in a single season, and Curry’s shooting is the primary reason. He has separated himself from the physical rules of basketball, to the point that distance from a defender and distance from the basket have become largely irrelevant. Curry is pushing the game towards change faster than anyone has in a long time.
Change can be scary.
I spent this weekend at the MIT Sloan Sports Analytics Conference in Boston, along with nearly 4,000 other attendees. Among those crowds were representatives from nearly every team in the four major sports leagues, all there for a slew of panels, product demonstrations, and research papers. The conference is where the sports analytics intelligentsia meet annually to discuss and plot the future of sports. This year, I attended presentations on biometric research that means to accurately predict athlete injuries before they happen. This is not theoretical either, the researcher claims to already be in the process of preventing injuries for clients.
A group of basketball researchers presented their work which used machine learning on raw SportVU data — this is the spatial tracking information captured by cameras in every NBA arena, 25 frames a second — to automatically identify when a screen had occurred and what sort of defensive strategy had been used to defend the screen. This might sound obtuse and somewhat absurd, humans already do this pretty well. But the work of this groupw allowed the aggregate analysis of a season’s worth of data, revealing that team are increasingly going over screens in an effort to prevent three-pointers. Their work also showed that Draymond Green and Curry were one of the most efficient pick-and-roll combinations in the league, regardless of the defensive strategy used.
If Curry and his three-point shooting represent the future of the NBA on the court, the research and discussion of the Sloan Conference represents the future of basketball off the court — new techniques for analyzing player and team strengths and weaknesses, new training techniques to preserve health and maximize physical performance. Some of this stuff sounds, admittedly, like science fiction. But this new future is not just a linear march forward. In fact, a lot of it seems to be circling right back to the beginning.
The NBA’s SportVU spatial tracking data has been the favorite data set of Sloan Conference researchers for several years, and even the sliver of data that is publicly available has provided a wealth of insight. Being able to separate shot attempts by pull-ups and catch-and-shoot, or to see how many times a player touches a ball in a game has added a fascinating level of granularity to our understanding of how offenses and defenses function. But we always knew these things existed, no new statistical species have been discovered. We’re simply able to measure things that we previously guessed at.
One of the most frequently ridiculed statistics in the public SportVU offerings is distance traveled. I mean, who cares if a player ran 3.5 miles or 4.1 miles over the course of a game? But in the Basketball Analytics: Hack-a-Stat panel, Mike Zarren, an assistant GM for the Boston Celtics, specifically singled out that statistic as something that had been revelatory for his organization when it was introduced a few years ago.
“Should you train to run three miles or five miles? Those are really different training regimes, right? We didn’t know until we installed the cameras during the playoffs in 2010, how far you guys were running during the games.”
In all of this talk about data, analytics, technology, and the future of the game, it’s important to remember the decidedly old-school thread that connects them all — the game of basketball. What Zarren is talking about is big data and big technology, filtering down to the simplest goals of a basketball practice — getting the players into game shape. The implementation of that knowledge is not space helmets and biometric muscle replacements, it is a trainer working with a player to make sure their body is up to the challenge of playing well for an entire game. The machine learned insights about screens and screening defense aren’t a precursor to robot coaches, they’re simply putting numbers to something that was previously limited by the ability of some poor video assistant to sit, watch, and tabulate by hand.
At another conference presentation, Kirk Goldsberry, Sloan veteran and newly minted Vice President for Strategic Research for the San Antonio Spurs, dealt with the growing chorus calling for the NBA to move back the three-point line. The challenge is both an aesthetic one (some people just don’t like threes) and a strategic one — a structural way to limit the absurd impact of Curry. Goldsberry offered some analytically derived ideas for redrawing the three-point line. His first was, instead of making the three-point line a static distance from the basket, to have it be a contour line that represents where the league average three-point percentage is 33.3 percent. This would eliminate any advantages from uneven geometry, like the corner three. His second idea was to let each team place the three-point line wherever they would like in their home arena (within reason), similar to how each baseball team has the freedom to set their own outfield dimensions.
Adjusting the court dimensions, be it moving the three-point line or widening the court itself, would be a dramatic step. It would not be unprecedented though — we can thank Wilt Chamberlain for the lane being four feet wider than it used to be. But by comparison, both ideas seemed absurd — which Goldsberry freely admitted in stating himself to be an avowed defender of Curry and the aesthetics of the Golden State Warriors. He was asked to give a presentation on creative solutions and so he did, even if he didn’t actually agree with the definition of the problem. At a conference, ostensibly about the future, here was another example of support being offered for elements of the past and present.
The NBA is going to continue to grow and change, it has for the entirety of its existence. But I call hyperbole on anyone who says it is headed for something unrecognizable. The future will probably have more three-pointers in it and it wouldn’t surprise me at all if the quest to find the next Stephen Curry consumed scouts, advertising dollars, and our imaginations — just like the quest for the next Jordan did. The players will hopefully be healthier and better informed about their own strengths and weaknesses, as well as those of their opponents. Coaches will have more information at their disposal but the challenge, just as it was 50 years ago, will still be to figure out how to get your guys to score more points than the other guys. Basketball played two decades from now will be more similar than it is different.
I saw the future, and it looks a lot like the present.