Turf Racing Analytics
Ranking algorithms for horse racing.
We deploy learning-to-rank models on historical turf databases to order runner win probabilities, delivering automated predictions.
Machine Learning R&D
Learning-to-Rank Predictions
Rather than running simple regression models, our turf prediction system implements an advanced spatial density ordering algorithm. Instead of attempting to forecast exact race times, the model learns to order the runners within a specific contest, reflecting competitor interactions.
- Recent Outings: Evaluating detailed historical metrics from the last 5 starts, margins, and non-finishing rates (DNF).
- Track Specifics: Win rates categorized by track surface (dirt vs turf), weather, and distance bands.
- Jockey & Trainer Influence: Statistical weights derived from jockey/trainer e-win rates over rolling 30 and 60-day windows.
Automated Ingestion Pipelines
Our database ingestion system scrapes up-to-date entry lists and weather changes from national tracks, recalculating runner scores and distributing updates automatically before each racecard.
TS