Model ops
did the model get better? — eval leaderboards, calibration and paper backtests
Prediction log — tamper-evident research record
Logged
0
predictions on record
Resolved
0
0 pending
Mean row Brier
—
resolved subset
Pick hit rate
—
argmax pick vs result
| Kickoff (UTC) | Match | Sport | Model version | p(H) | p(D) | p(A) | Pick | Status | Result | Row Brier |
|---|---|---|---|---|---|---|---|---|---|---|
Row Brier is the multiclass sum convention (uniform 3-way scores 0.6667; red rows scored worse than uniform). Two-way sports carry no draw column. Pending rows resolve when the result lands; the log is append-only in the real system — resolution never rewrites the forecast.