Modeling NBA Defensive Rotations
Our research models basketball team dynamics as active matter using Wasserstein Gradient Flows. By applying the JKO scheme, we simulate how defensive densities evolve to minimize a potential energy functional, responding to offensive threats and defensive formation in real-time.
Defensive densities “flow” across the court to minimize the energy functional, reacting dynamically to offensive spacing and ball movement.
Using Bayesian Optimization, our model learns the hyperparameters of the potential functions from historical tracking data.
This visualization tracks the gradient flow surface as defenders adjust their positioning based on the ball and offensive threat levels.
Defensive Modeling Results
This project provides a novel framework for analyzing defensive efficiency and spatial optimization through the lens of fluid dynamics and optimal transport. Explore our full gallery of outcomes, including spatial metric-measure structures and topological defensive identities.