Player motion times predicted by quantile regression outperform physics models
Learning Human Sports Motion Feasibility via Quantile Regression on Tracking Data
Artificial Intelligence
Summary
This paper looks at how quickly sports players can reach different spots on the field based on their current movement. Instead of using traditional physics or guesswork, the authors teach a computer model to predict the fastest possible time a player could get somewhere by learning from real tracking data. They tested their method on soccer games from the 2022 World Cup and found it worked better than existing models. They also checked if the model could work for basketball and American football data, exploring if these motion patterns transfer across sports.
What this means in practice
- •For sports analytics teams: Estimate the quickest a player can move to a location using learned models instead of traditional physics formulas, improving tactic and performance analysis.
- •For video game developers: Create more realistic player movement timings in sports games by integrating data-driven prediction models of player locomotion.$Commercial implications: Enables more accurate and responsive player motion in sports video games, enhancing gameplay realism based on real-world movement patterns.
Authors
Marc Gutiérrez-Pérez, Antonio Agudo
Abstract
We study time-to-arrival (TTA) in sports: the shortest plausible time a player needs to reach an arbitrary location on the pitch given their current kinematic state, estimated as the lower envelope of observed locomotion times. Rather than modeling space ownership through position density estimation or closed-form physics-based models, we cast TTA as a conditional quantile regression problem, estimating the lower quantile of locomotion time so the model learns a player's lower-envelope arrival time. We train the model on optical tracking data from all 64 matches of the PFF FC FIFA World Cup 2022 dataset, and evaluate a lightweight multilayer perceptron against six player-motion models from the literature, re-expressed as TTA predictors, and a learned reachable- hull baseline. Across our evaluation metrics, the MLP consistently outperforms every physics-based baseline. To probe the generality of the learned locomotion model, we further evaluate it on basketball tracking data and a single-game NFL tracking dataset, analyzing the transferability of player motion dynamics across sports. Our project is available at https://github.com/mguti97/TTA-Quantile.