SMOGS: Social Network Metrics of Game Success

TL;DR

This paper introduces a novel social network-based metric for predicting basketball game outcomes by modeling player interactions and passing patterns using advanced stochastic processes and latent factors.

Contribution

It extends stochastic process modeling to dynamic basketball networks, incorporating latent factors to distinguish successful games from losses.

Findings

Latent factors differ significantly between wins and losses.

The model accurately recovers parameters in simulation experiments.

Applied to high-resolution tracking data, it reveals key interaction patterns.

Abstract

This paper develops metrics from a social network perspective that are directly translatable to the outcome of a basketball game. We extend a state-of-the-art multi-resolution stochastic process approach to modeling basketball by modeling passes between teammates as directed dynamic relational links on a network and introduce multiplicative latent factors to study higher-order patterns in players' interactions that distinguish a successful game from a loss. Parameters are estimated using a Markov chain Monte Carlo sampler. Results in simulation experiments suggest that the sampling scheme is effective in recovering the parameters. We then apply the model to the first high-resolution optical tracking dataset collected in college basketball games. The learned latent factors demonstrate significant differences between players' passing and receiving tendencies in a loss than those in a win. The model is applicable to team sports other than basketball, as well as other time-varying network observations.