Keywords
Racing, statistical analysis, modelling

Technologies
Python, R, Power BI, Pandas, Seaborn

National Hunt racing presents inherent welfare risks, as horses must jump obstacles at speed across the full race distance. Non-completion is a key indicator of safety, and understanding what drives it is essential for evidence-based policy.

This project analyses non-completion rates across Great Britain, Ireland, and France using a dataset of over 11,000 races and 110,000 race entries, scraped from Sporting Life and rigorously cleaned to ensure cross-country consistency. Exploratory analysis examined how factors such as horse age, ground conditions, and race type relate to non-completion risk.

Logistic regression models were built at both the horse and race level to quantify these relationships. The models revealed that France has the highest non-completion risk of the three countries. Younger horses, extreme ground conditions, and steeplechase races were all associated with elevated risk.

These findings offer racing authorities and welfare bodies a clear, interpretable evidence base for targeting the most modifiable risk factors. The project also points toward future work, such as developing tools to flag high-risk race conditions before they occur.