Open-Source Hong Kong Horse Racing ML Pipeline — Feedback Welcome [P]
A developer has created an open-source machine learning pipeline for predicting Hong Kong horse racing outcomes, focusing on building a reproducible system rather than claiming AI superiority. The pipeline includes feature engineering, various model training options with and without betting odds, ensemble predictions, and simulation of different betting strategies. A key finding is that the model without odds outperformed the one with odds for quinella bets, suggesting that public odds efficiently price favorites but may miss mispriced combinations. AI
IMPACT Provides a reproducible ML pipeline for a niche domain, potentially useful for other prediction tasks.