Win % = model's chance the horse wins Β·
Place % = chance of finishing top 3 Β·
Sharp = our highest-confidence niche (32-35% win rate historically)
Win %
Place %
Top-2 gap
Win % gap between the model's rank-1 and rank-2 horse. Bigger gap = clearer favourite. 60-day backtest: only β₯20pt was historically +EV at flat stake (33.7% win, +2% ROI).
π Value
Picks where the model gives the rank-1 horse at least 5pt more win chance than the market implies, AND the bookmaker is offering $4β$10. 180-day backtest: 220 races, 19.1% win rate, $5.48 avg winner SP, +32.7% ROI at flat $10 stake. Mid-priced overlays β the model thinks the market has them too long.