How Confidence % Is Calculated
Each pick's confidence comes from an 11-factor weighted model that analyzes recent form (L10), home/away splits, head-to-head history, schedule fatigue & travel, pace, sportsbook odds, prediction markets, star player impact, injuries, offensive/defensive efficiency, and position matchups.
Spread & ML: The 11 factors produce a composite score that's converted to a win probability via logistic regression. The predicted spread is compared to the book's line — the gap (edge) is converted to confidence using 1/(1+e-edge×0.42), capped at 50–95%.
Over/Under: Both teams' recent scoring averages (points for + against) predict the game total. The gap between our predicted total and the book's line drives confidence through the same logistic function.
Ranking Score = Confidence + EV Bonus (+EV × 0.5, max 15) + Edge Bonus (edge × 1.5). Picks with higher confidence, larger edges, and positive expected value rank higher.