March is bound to get even madder.
Scientists at Cornell University have put together a data model that suggests that the application of a physics theory to basketball may lead to teams scoring five to 15 more points per game.
Researchers analyzed player metrics and material that were accrued from an undisclosed NBA team through a stop-motion camera during many of its games this season. The science squad was then able to project precise positioning that guaranteed better scoring outcomes for individual players — sometimes by moving mere inches.
“Every 40 milliseconds, we know with … a very high degree of accuracy, where every player is and where the ball is located,” Boris Barron, a doctoral physics student on the project, told The Post.
“[Our work] has the potential to be a game changer for basketball … This is taking ‘Moneyball’ to the extreme.”
Although the Big Red missed the big dance, Barron — along with physics professor Tomás Arias and peer Nathan Sitaraman — have been on their toes these past few weeks by applying density-functional fluctuation theory (DFFT) to introduce “more kind of advanced quantitative analysis” to the game.
In quite plain terms, DFFT looks at fluctuations caused by certain events that either separated or brought together entities within a group. Previous research using the theory observed how fruit fly clusters adapted to heat being introduced to their environment and separately, was used to predict crowd behavior among people.
Barron and company are using DFFT to break down the spatial interactions of where players like to be and how players interact with one another on the court.
“Looking back at a game, I can see how this can help players improve,” Barron said. “The improvements can be in the [team total] range of five points in 100. It wouldn’t shock me based on the results that we’re getting here,” he added, mentioning that there could “potentially” be upticks by 15 points or more.
The approach can quantify a player’s success, or lack thereof, from several nearby positions on the court — thus predicting more exact locations where they will score more or defend better in just about any given scenario.
“We can take a look at a snapshot of a game and ask, does this look like a good position for the offense? Or does this look like a bad position for the offense?” Barron said.
“Where this becomes useful is that we can improve a player’s positioning,” he added of the data, which currently only accounts for two-point shots.
Former Oakland A’s general manager Billy Beane found incredible success with another data intensive strategy — “Moneyball” — in the early 2000s.
Beane was constantly asking “but can he get on base?”
In that same vein, many basketball coaches may soon pose the question “but can he drive to the net?” from simulations based on the Cornell research.
“We’re determining where each of the players should move,” Barron said. “We’re pretty much saying ‘this guy, in this case, should prefer to take kind of this path [to the basket].’ “
Statistics wrung from DFFT simulations can hyper-analyze positioning to help teams better scout future opponents and individual matchups.
Admittedly, more variables — like accounting for players’ set positions, specialty skill sets and re-running the numbers to include three-pointers — still need to get worked in, according to the doctoral student.
“Maybe [next] we can follow along a certain kind of player and see if they tend to stand in good positions for the team or maybe not so good positions for the team,” he said.
“You can imagine turning some of our modeling into a simulation tool for coaches.”
Even with changes to come, Barron said the theory behind what they’re shooting for is sound at the moment.
“Going forward, you can imagine using this to provide a positioning metric for basketball.”