Project: The Rugby League Integrated Study into Game and Training demands; RL-InSiGT project
Causal modelling of player performance, injury risk, and head accelerations in professional rugby league
The project will develop and apply Directed Acyclic Graphs (DAGs) and advanced causal analysis to understand how a) law modifications, b) tactical changes or c) changes in training exposure influence, key metrics measured via Catapult player tracking units, in addition to other outcome measures. These include:
Player physical performance
Players and team match events
Concussion risk
Head acceleration event risk
You will work with large, high-resolution, multimodal datasets, including:
GPS and inertial data from player tracking systems
Match event and contact data
Instrumented mouthguard head-impact data
Longitudinal injury surveillance records
The research is conducted in collaboration with the Rugby Football League, providing access to rare, league-wide datasets and direct pathways to real-world impact.
This PhD is methodologically driven with a focus on the real-world application. While the application domain is men’s rugby league (Super League), the core intellectual challenge lies in causal structure learning, confounding control, missing data, measurement error, and decision-relevant inference in observational settings.
WHY CAUSAL INFERENCE?
Much applied analytics in sport, as in many other fields, focuses on prediction: forecasting injury, performance, or fatigue from historical data. However, prediction alone cannot answer the questions that matter most for real-world decision-making:
What would happen if variables were changed?
Which variables increase concussion risk or performance outcomes, rather than merely correlate with them?
How do interventions trade off performance gains against risks?
These are causal questions, not predictive ones.
This PhD is explicitly built around causal inference as the primary analytical framework. Professional rugby presents a rare and intellectually rich environment for causal analysis: repeated interventions, time-varying confounding, feedback loops, missing data, and noisy measurement systems, all observed longitudinally in a real-world setting.
The project treats causal structure as a first-class object of study, using:
Directed Acyclic Graphs (DAGs) to formalise assumptions, identify confounding, and define valid adjustment sets
Counterfactual reasoning to estimate the effects of hypothetical changes in training, exposure, or match demands
Longitudinal causal models to address cumulative exposure and time-dependent confounding
Transparent and interpretable models designed to support decision-making, rather than black-box prediction
The aim is not simply to model rugby, but to generate robust, decision-relevant causal insights from complex observational data — skills that transfer directly to fields such as health, epidemiology, safety science, and human performance analytics.
RESEARCH AIMS
The PhD will aim to:
Construct and validate causal graphs describing relationships between match events, physical outputs, player and team performance, concussive and head acceleration event outcomes
Apply causal inference techniques (e.g. adjustment sets, mediation analysis, counterfactual estimation) to quantify risk and performance trade-offs
Integrate heterogeneous data sources with differing temporal resolutions and noise characteristics
Develop interpretable, decision-support models for performance optimisation and player welfare
The work will contribute both to applied sport policy and to methodological discussions around causal modelling in complex, real-world systems.
YOUR OPPORTUNITY
As the successful candidate, you will:
Work with large-scale longitudinal datasets rarely available
Develop causal models with direct relevance to critical decision-making, balancing the safety vs. the spectacle of the sport
Collaborate with academic researchers, data scientists, and industry partners
Publish in high-quality peer-reviewed journals at the intersection of statistics, data science, and applied health/performance research
You will be supported to build strong transferable skills in causal analysis, statistical computing, and applied machine learning, with clear relevance beyond sport (e.g. health, epidemiology, human performance, and safety analytics).
In addition to the PhD, you will have the opportunity to provide real-world data insight back to the sport.
Applicants are encouraged to discuss their proposals with a member of the supervisory team: Professor Ben Jones, Dr Thomas Sawczuk, Dr Cameron Owen, Dr Neil Collins or Professor Mark Gilthorpe (Advisor)
