Sports Performance Data Scientist

Anton Volkov

Freelance · Mid-level · London

This role is no longer accepting applications

Similar open roles you can apply to now:

About the role

We are the performance analytics department of a professional rugby union club. We compete in the Premiership. We have a head of performance, two sports scientists, a video analyst, and until recently no one who could do serious data modelling. That has changed because our data situation has changed: we now collect GPS tracking data from every training session and match, accelerometry from wearable sensors, HRV and sleep data from player wellness apps, and historical injury records going back six years. We have never used any of it in a statistically rigorous way. Our sports scientists are excellent practitioners. They understand rugby and athlete physiology. They do not know how to build a predictive injury risk model or how to identify which GPS load metrics are actually predictive of match performance decline in the following week. We are commissioning a 12-week project to produce two things: a training load monitoring model that identifies players at elevated injury risk based on GPS and wellness data, and an analysis of which performance metrics from GPS tracking are most predictive of subjective coaching ratings of match performance. These are exploratory analyses, not production systems. We want rigorous methodology, honest uncertainty quantification, and findings we can act on in our daily training decisions.

Key Deliverables

List the expected deliverables for this project.

Clean and integrate GPS tracking, wellness, and injury record datasets for the project scope — 2 seasons of first-team squad data
Build and validate a training load-to-injury risk model, with clear documentation of assumptions, limitations, and confidence intervals
Conduct feature importance analysis identifying GPS metrics most predictive of performance decline
Produce a written analysis report and presentation deck suitable for delivery to the head coach and performance director
Conduct a handover session with the sports science team including guidance on how to update the models with new season data

Requirements

Technical stack needed for this mission.

3+ years of data science with experience in sports analytics, physiological data, or time-series prediction in applied contexts
Python at a professional level — clean, reproducible analysis code with documented methodology
Statistics at depth: mixed-effects models, survival analysis for injury risk, and the intellectual honesty to distinguish real signals from overfitting in small sports datasets
Pandas and NumPy for time-series data manipulation from GPS and wearable sensor systems
Scikit-learn for predictive modelling, cross-validation, and feature importance analysis
SQL for extracting data from our athlete management and GPS platform databases
Familiarity with the statistical challenges of small-N sports datasets — underpowered studies, individual variability, and the limits of what the data can support