Open, transparent, data-driven analytics for Turkish horse racing. 40.8% Top-1 accuracy.
Quiet but deep.
Open feature set
Every feature the model sees — track, distance, jockey, trainer — listed in plain view.
Decision rationale
Each prediction shows its top 3 factors and weights — no hidden black box.
Historical archive
13,587 races, 130,930 runners — including where the model has been wrong.
Backtest panel
Track the model's day-by-day performance against the past 12 months of real results.
Discipline tools
Not betting — educational tools for budget management, variance and stop-rules.
Explainability
Why this runner, why now — SHAP explainability translated into race-specific language.
An open ecosystem that turns horse racing from emotional gambling into an auditable quantitative discipline.
Present sports data with the transparency it deserves — without abandoning intuition, just placing data alongside it as an analyst.