Advanced statistics are a lie.
Points per possession. Expected field goal percentage. Rebound rate. They’re all incredibly useful tools in breaking down who does what on a basketball court. They’re several orders of magnitude more informative than simple per-game metrics. They’re the foundation upon which our current understanding of basketball is based.
And they’re a first-grade remedial math class in relation to the true advanced analysis that’s on the horizon.
As the MIT Sports Sloan Analytics Conference got into the meat of its presentations, it became increasingly evident that our pace- and opportunity-adjusted numbers are only scratching the surface of the true nature of the game. If what we know is an elevator, what we’re using now as “advanced stats” are the lobby. The SportVU optical tracking data that more and more teams are using (and isn’t available in the public sector, save for a few tantalizing glimpses here and there) is the second floor.*
*Or first floor, for you crazy non-Americans. Seriously, how does that even make sense?
The penthouse level — the true advanced measurements that represent that ever-nebulous concept of “the future” — actually has very little to do with what happens on the court, in a sports sense. Instead, “The Science of Team Chemistry” is the foyer to our deluxe apartment in the sky. As fancy and informative as our numbers are, what really matters is the interaction of human beings on a lacquered floor made of wood. We know this, on an intrinsic level — even those of us who pray at the altar of advanced statistics. Team chemistry and interpersonal dynamics matter. The problem to this point has been how we determine how much they matter. Evaluating whether a player wants to be teammates with Kobe Bryant, and how effective he is when teamed with Kobe has been dependent on abstractions from on/off court data, adjusted plus-minus numbers and the good old eye test.
These are the limitations of basketball statistics, no matter how precise they are. To go beyond those limitations, we must reach out to other disciplines. Daniel McCaffrey and Kevin Bickart come from a neuroscience background — an environment in which experiments based on chemistry are the baseline, not the exception. The most famous experiment they cited studied the relationship between heart rate synchronicity (that is, how similar were the heart rates) in married couples and their ability to cooperate to solve a puzzle. Couples whose heart rates were most in sync experienced lower levels of interpersonal conflict, solved the puzzle faster and, perhaps most fascinatingly, were more likely to still be married in the future.
Working from this laboratory result, our presenters looked at heart rate data gathered during games and created some of the most beautiful motion charts I’ve ever seen to illustrate an out-of-bounds throw-in to the box in a women’s soccer game. The motion chart displayed heart rate synchronicity in space. Those players whose heart rates were most similar congregated to the center of the chart, while those whose physiological reactions were less related floated around the perimeter, though still connected to the other nodes by lines that represented some level of similarity. Finally, a player completely removed from the play appeared at the top, like a lonely Pluto pondering its cosmic fate.
Before the play begins, the players on the pitch — represented by simple nodes — are very loosely connected in their heart rate synchronicity. When the ball is entered and the action begins, though, an absolutely fascinating image went into motion. The nodes for those players whom shared the burden of play (passes, the shot, creating proper space in the box) and did so well were practically on top of each other. This didn’t necessarily relate to spatial proximity on the pitch; for instance, the player throwing the ball in immediately synchronized heart rate with the player receiving the ball. Their performance connection went deeper than a simple pass. At the most fundamental level, they were displaying what happens when team chemistry manifests itself during the course of competition.
Call it an intangible at your own risk, for these are creatures of flesh and blood. And that’s the most important takeaway from the analytics of chemistry: we are dealing with people. As much as we want to — and can — reduce a person’s performance to a set of numbers based on results, doing so confuses the effect for the cause. A Chris Paul to Blake Griffin halfcourt alley-oop is infinitely entertaining and awesomely effective, but it’s not effective simply because those are two very good players. It’s highly likely that their productivity on the court is a result of the interpersonal dynamics of two human beings who really like playing basketball together. “The Science of Team Chemistry” aims to lead us down the road to defining just how likely that is and how much that connection matters.
For all of the specter of the presentation, it was not without issues. Foremost is the problem of gathering the data; so far, the only way to measure heart rate during games is with cumbersome, uncomfortable chest straps.* Teams and players are understandably unwilling to compromise performance for a bit of biological data, creating a bit of a chemistry Heisenberg principle: does measuring performance mean measuring sub-optimal performance? Secondly, player privacy is a real concern. McCaffrey said that one of the biggest issues that they face is convincing teams to allow them to work with the players. Beyond that, this new information represents a broadening of the health/injury portfolio of a player, as it were. If we extend this type of testing to its logical outcome, the new frontier of analytics will be brain chemistry, brain types, pain tolerance, susceptibility to “checking out” — all things that could have a significant impact on a player’s earning potential. Would a team attempting to retain an unrestricted free agent leak information indicating that a player has certain mental characteristics that make them difficult to work with? McCaffrey made it clear that they operate under the strictest confidentiality guidelines (given their background in health-related fields), but that’s not to say that the people at the very top of an organization would use the information with such discretion.
*We’re talking about straps so restrictive that football players, even with all of the equipment they typically wear, found them so limiting that they’d simply rip them off of their bodies at the earliest opportunity, per McCaffrey.
Lastly, the data itself left something to be desired. How much of the heart rate synchronicity was correlation caused by oxygen use and energy expenditure? Is this type of measurement more indicative of similar levels of effort? Is what we know from a laboratory setting about two hearts beating as one and its effect on team performance applicable to a free flowing competition?
Yet for all of those questions surrounding this particular dataset, the specific answers aren’t what’s important. Even if this specific research is only marginally related to the NBA — and I believe it’s much more important that that, but even in the worst case — it represents a groundbreaking view of our beautiful game. What matters in basketball isn’t just points, be it per game or per possession. The true underpinnings of the game are the people who play it. The better we can evaluate those people, as human beings and not just commodities who perform according to a model, the better our understanding and analysis.