Strategy is widely recognised as a critical factor in track cycling sprint performance. Yet measuring strategy using quantitative evidence remains a significant challenge. To overcome this hurdle, we adopted a data-driven approach by analysing videos of 1,500 matches in World and European championships, World Cup and Olympic Games from 2018 to 2024. By processing videos, we visually identified the rider`s respective position at key distances. Then we developed a machine learning model to predict race outcomes. Results show that if starting position has no impact on the race outcome, match configuration can drastically change a rider`s chances of victory. Then, we evaluate the ability of international riders to win matches depending on the opponent skill. Top riders achieve a 90% overall victory rate while frequently defeating faster opponents. Then we use these results to build a model that uses these results to predict race outcome with 83% accuracy. Finally, we apply this model to Paris 2024 Olympic Games sprint competitions. Overall results highlight key factors that impact victory likelihood in sprint; rider mean velocity in 200 m time trial, tactical skills and race configuration in decreasing importance.
© Copyright 2025 International Journal of Performance Analysis in Sport. Taylor & Francis. All rights reserved.