Runner comparison

Active field and statistical profile

Requires data
  • Form, weight, barrier, jockey and trainer
  • Distance, course and condition suitability
  • Scratches trigger field-level recalculation

Probability distribution

Mathematically coherent race-level estimates

Requires data
  • Active-runner win probabilities sum to 100%
  • Place estimates respect field size and paid places
  • Market baseline remains separate from model score

Explainable forecast

Verified feature effects, not invented commentary

Requires data
  • Supporting and risk factors
  • SHAP contribution boundary
  • Feature cutoff and model version audit