Keywords
Horse Racing, Statistics, Scraping, Welfare, Data

Technologies
Python, R, R Shiny, Selenium, Beautiful Soup

Non-completion outcomes are a key indicator of safety and welfare in National Hunt racing. This study presents the first cross-jurisdictional analysis of safety-related non-completion across the UK, Ireland, and France during the 2022-2023 jump racing years. An automated scraping pipeline assembled a dataset of 112,367 horse-starts across 12,397 races, which was cleaned and standardised across jurisdictions.

A horse-level logistic regression model and a race-level quasibinomial model were fitted, incorporating country-by-covariate interactions to assess whether risk factors differed across jurisdictions. Substantial jurisdiction-specific differences were observed. Most notably, Fence races carried elevated non-completion odds in all three countries, but the effect was greatest in Ireland. Distance effects were strongest in France and largely absent in the UK, while Weight showed a protective association in France that was absent elsewhere.

These findings demonstrate that non-completion risk reflects jurisdiction-specific differences in race format and course design, supporting the need for tailored welfare and regulatory strategies across European National Hunt racing.