Every pick we've called, graded against the final score.
No survivor bias, no deletions. The SportsSlateAI model's full track record — unit-weighted ROI that scales with conviction, win/loss breakdown, and every pick resolved or pending. Updated as games finalize.
Every pick is captured at publish time with the odds, edge, and conviction units we staked. After the game finalizes we grade against the actual final score — no retrospective adjustments, no swapping sides after the fact. Picks are written to pick_results once the article publishes and stay there whether they win or lose.
Unit-weighted ROI reflects the model's conviction — higher-edge plays risk more units. Stake sizing is continuous: units = min(edge% × 0.5, 10). A 6.3% edge stakes 3.15u; a 10% edge stakes 5u; a 20%+ edge tops out at the 10u cap. A winning play pays units × (100 − cents) / cents; a loss returns −units. Any edge below 3% is a 0-unit pass — recorded and graded for calibration but not staked, so it's excluded from the ROI denominator. Pushes and voids are neutral and also excluded from the denominator.
1-800-GAMBLER · Must be 21+ · Prices via Polymarket + Kalshi · Directional leans only, every number validated.