Basketball Player Tracking Data for Expected Possession Value Modeling

A framework for estimating expected possession value (EPV) in basketball using optical player tracking data, proposed by Daniel Cervone of New York University. The model is a multiresolution stochastic process differentiating between continuous player movements and discrete events like shots. A data sample and R code are provided in the supplementary material for model exploration.

Use Cases

• Predicting the expected points from a basketball possession in real time based on player tracking data.
• Analyzing player decision-making tendencies as a function of spatial strategy using the hierarchical model.
• Estimating transition kernels for possession events using hierarchical spatiotemporal models on large datasets.

Strengths

• Model is computationally tractable on very large datasets, as stated in the description.
• Framework provides a data sample and R code for further exploration, indicating some level of reproducibility.

Limitations

• Row count is unknown, which may limit suitability assessment.
• Column-level documentation is absent; field semantics must be inferred after download.
• Last update date is unknown; freshness unverified.

Provenance

Source

Daniel Cervone, New York University

Collection Method

Likely derived from optical player tracking data, though the specific collection method is not detailed.