Publications
Inferring Player Location in Sports Matches: Multi-Agent Spatial Imputation from Limited Observations Link to heading
Citation: Everett, G., Beal, R.J., Matthews, T., Early, J., Norman, T.J. and Ramchurn, S.D., 2023, May. Inferring Player Location in Sports Matches: Multi-Agent Spatial Imputation from Limited Observations. In Proceedings of the 2023 International Conference on Autonomous Agents and Multiagent Systems (pp. 1643-1651).
Description: Understanding agent behavior in Multi-Agent Systems (MAS) is critical in fields like autonomous driving, disaster response, and sports analytics. This work tackles agent location imputation in environments with non-uniform timesteps and ~95% missing values. Using Long Short-Term Memory and Graph Neural Networks, we predict agent locations, applying this to football by imputing player locations from sparse event data. Our model increases accessibility to analysis such as physical metrics and pitch control without the need for costly optical tracking data.
The Strain of Success: A Predictive Model for Injury Risk Mitigation and Team Success in Soccer Link to heading
Citation: Everett, G., Beal, R., Matthews, T., Norman, T. and Ramchurn, G., 2024. The Strain of Success: A Predictive Model for Injury Risk Mitigation and Team Success in Soccer. In 18th Annual MIT Sloan Sports Analytics Conference 2024.
Description: Player injuries in soccer significantly impact team performance, club financial stability and player welfare, with the ‘Big Five’ European soccer leagues experiencing a staggering £513 million in injury-related costs during the 2021/22 season. In this paper, we present a novel forward-looking team selection model, framed as a Markov decision process and optimised with Monte Carlo tree search, that balances team performance with the risk of long-term player unavailability due to injury. We demonstrate that real-world teams could reduce the incidence of player injury by ~13% and wages inefficiently spent on injured players by ~11% using our data-driven team selection model.
Contextual Expected Threat using Spatial Event Data Link to heading
Citation: Everett, G., Beal, R., Matthews, T., Norman, T. and Ramchurn, G., 2022. Contextual Expected Threat using Spatial Event Data. In Statsbomb conference proceedings.
Description: We present a spatial Expected Threat (xT) model that computes goal likelihood using spatial event data. Using this spatial xT model, we propose a novel way of valuing attacking and defensive units in football by computing a ‘Threat Above Expected’ (TAx) metric which measures the change in goal probability resulting specifically from spatial context. Using this metric, we evaluate team defences and attacks, and demonstrate a concept for a positional defensive optimiser that can offer coachable insights by identifying improvements in defensive setups during post-match analysis.