Verification
Model accuracy tracking begins after the first completed match. Until then, this page shows metric definitions and methodology.
Status
pre_tournament_placeholderpre_tournament
Metric definitions
- Brier
- Measures the mean squared difference between predicted probabilities and actual outcomes. Lower is better. Range: 0 (perfect) to 1 (worst).
- Log-loss
- Penalizes confident wrong predictions more heavily. Lower is better.
- Calibration
- Measures whether predicted probabilities match observed frequencies across buckets.