Edge vs Actual Outcome

Measures whether the model's edge predictions match reality. If model predicts +20% edge on a pick, that pick should win 20% more often than its Kalshi price suggests. Drift from diagonal = miscalibration.