Nylon Notebook: Nylon Calculus at the 2016 MIT Sloan Sports Analytics Conference
Nylon Calculus is well-represented at this year’s MIT Sloan Sports Analytics Conference and we’ll be bringing you our thoughts and reactions throughout the conference. Check back in here during the day for updates.
The Moneyball Generation
This year’s Sloan Sports Analytics Conference began in earnest with a Moneyball reunion panel. Which is fitting, right? When most people think of sports analytics, they think of Moneyball. And this, as the 10th Sloan Conference, made perfect sense to bring back some of the main folks behind the book (notably, sans Billy Beane).
Moneyball was released way, way back in 2003. The movie came out in 2011. To some, that might feel like not that long ago. But at this conference, where younger and younger individuals show up each year — no doubt college and high school students make up a large portion of this year’s 3,900-plus person record crowd — 2003 is somewhat of ancient history.
Sports analytics has been around for decades. Bill James, one of the fitting members of this first panel, had been publishing his almanacs for oodles of years before Michael Lewis ever conceived of Moneyball. But forever onward, Moneyball is the seminal moment of when sports and data reached the big time. This is when it became part of the usual public discourse of sports and commonly accepted by a growing part of the masses.
There is obviously still much more work to be done. Sports analytics has the potential to continue to be more heavily integrated into the public discourse. Teams and companies continue to do phenomenal behind the scenes. But what’s interesting to me as Sloan starts up again is how Moneyball really has changed the sports fandom experience for young people today.
I really enjoyed this article from The Wall Street Journal’s Ben Cohen last April. He wrote about middle school students (!) having detailed conversations about shot selection, inefficient long twos, and tanking. What were you talking about in reference to sports when you were just 12 or 13? When I was that age — and yes, this dates me — Moneyball was first released. For middle schoolers today, Moneyball has been around their whole lives. It’s become normal and ingrained in the day-to-day conversations that they have.
The sports fans, sports writers, and sports decision-makers of tomorrow will just be used to this world of the intersection of sports and data. There’s no fighting it; it’s just a way of life. Over time, this whole us vs. them dilemma will finally fade away because all young people will just be comfortable with the concept of Moneyball and the use of analytics in sports. The discourse will continue to evolve and evolve from there.
— Jacob Rosen, @JacobLRosen
Make The Machines Do It
There is an increasingly realistic vision of the future where smart machines assume responsibility for all the tedious and mundane tasks that humans no longer want to do. The efficiency of the machines liberates humans for meaningful and creative work. In basketball analytics, one task being slowly assigned to the machines is identifying screens. The pick-and-roll is an integral part of almost every NBA offense. Pulling useful insights about the pick-and-roll out of any data set necessitates identifying when a screen has occurred. Two years ago at the Sloan Conference, a paper was presented that used machine learning to identify ball screens. This year, Avery McIntyre, Joel Brooks, John Guttag, and Jenna Wiens presented their work which used machine learning to not only identify screens from spatial data sets, but also classify the defensive scheme used to defend the screen — specifically over, under, and switch.
It’s a fascinating project with near limitless applications.
The group showed that across the entire league more and more ball screens are being defended with the ball-handler’s man fighting over the screen. This is the defensive strategy with the best chance of limiting a three-pointer, and it would make sense that this would be an increasingly prevalent strategy as defenses try to keep up with skyrocketing three-point rates around the league. They also showed that Stephen Curry and Draymond Green (and then Curry and any other big) were extremely likely to be defended by fighting over the screen — and that they were incredibly efficient regardless of the defensive strategy.
We don’t need a machine to tell us that a Draymond Green – Stephen Curry is incredibly difficult to guard. But they might be able to help us tell which defensive strategies are (marginally) more effective. What an age we live in.
— Ian Levy (@HickoryHigh)
Mapping Basketball
Kirk Goldsberry is pretty much synonymous with the Sloan Sports Analytics Conference. His research papers and presentations have been a constant feature of the past few years and, of all the things he’s contributed to the basketball analytics conversation his persistence on putting analytics into the context of a three-dimensional spatial environment has definitely pushed things in new directions. Today, at the conference he gave a presentation sharing ideas about moving the three-point line.
Some fans and analysts have called for adjusting the distance of the line to equalize the impact of great shooters. Goldsberry wasn’t advocating for any particular solution (or the existence of a problem) but he acknowledged the impact of the line saying, “That little ribbon on the perimeter is the biggest economic factor in the behavior of players.”
One of the ideas he presented, was to think about the floor spatially and place the three-point line based on shot data — essentially, making the three-point line into a two-dimensional contour line that marks the distance where the league average field goal percentage would be 33 percent. This number equates to a 50 percent two-point percentage and would theoretically be far enough to discourage bad shooters without eliminating the fundamental advantage of the three-pointer for good shooters.
Another idea was to allow each team to, within a narrow range of options, place the three-point line wherever they would like. Teams could place the line where it benefits their specific roster, and increase the advantages of thoughtful, well-constructed team. It’s an idea that seems absurd but it’s a fairly close analogue to the variations in outfield dimensions from team to team in baseball.
Neither is likely to be adopted, but thinking about the three-point line as a function of space and just one strand in a web of advantages and disadvantages was an interesting addition to the current dichotomous shouting match of:
“move it back!”
“no, it’s fine!”
— Ian Levy (@HickoryHigh)
The Future of Biometrics Data in the NBA
Andy Glockner, friend of the site and editor at The Cauldron, just held a short discussion on the topic of basketball biometrics data. We seem to be on the cusp of some crazy cool and innovative technologies … but there is no knowing what really could come next.
Specifically, what is most interesting to me is where this could evolve due to the 2017 CBA negotiations. Ethically, who owns this data? Is it kosher for teams to be tracking players 24/7 (besides regular season games, because that’s not allowed) and use that data against them in potential negotiations?
Obviously, there still remain concerns over the margin of error and the types of actionable insights that are actually possible from tracking health data and player movement at all times. But is it worthwhile for the players to bring it up at the negotiating table next summer? There already exist concerns over how to account exactly for Basketball-Related Income. Players could bring up the topic of maximum salaries or the rookie wage scale or other topics related to wages. Bringing up something else could lead to potential concessions they aren’t willing to give.
If a team finds out something about a player that is concerning long-term and hurts that player’s potential, what should they do next? Should the player’s agent know this information? What about other teams? It’s a really fascinating continuing debate at the forefront of the sports analytics world. The regular biometrics-related articles from Pablo Torre and Tom Haberstroh are really, really interesting as a fan … but what will it look like in another few years?
— Jacob Rosen, @JacobLRosen