Every match. Every call. Stats. Stars.
The full ledger — every match in the 2026 World Cup, both tracks' predictions, both reasoning lines, the actual result. No edits after kickoff. Brier score at the top tracks each model's running calibration. Lower is better; 0.67 is the uninformative baseline.
Running Brier score
—
Stats (mean Brier)
Stars (mean Brier)
Lower = better calibration
Not enough graded matches yet — Brier curve will fill in as matches finish.
Loading match ledger…