Published
5
Every settled matchday is disclosed after the evaluation worker finishes: shortlist wins, misses, abstentions, league splits, and every modelled market. No outcome is hidden by a subscription.
This report is rebuilt from the evaluation worker's settled postmortems. It measures the exact market shortlisted for each match and separately audits every other modelled angle.
Saturday, 24 October 2026 · Lagos time (WAT)
The page never substitutes demo results. A report appears after completed fixtures have been processed by the evaluation worker.
This section remains separate from the reliability shortlist. It evaluates the original EPL 1X2 probability distribution, scoring quality, and publication-time odds.
Accuracy is wins divided by settled wins plus losses. Pending and void results are excluded. Brier score and log loss evaluate the original probability distribution; lower is better. The calibration gap compares average recorded confidence with observed accuracy. ROI and unit profit/loss are calculated only when every eligible settled record has immutable publication-time decimal odds; later prices are never substituted.
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Settled goal picks only, every league combined, each market with its own hit rate and sample. A two-way goal market is easier to hit than a three-way match result, so these figures are never added to the 1X2 record above. Every settled pick counts, losses included; voids and no-pick matches are excluded.
Stated probability against what happened, over the latest 22 settled picks:
Stated probability against what happened, over all 39 settled picks:
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·
Changed: system reforecast — Superseded before kickoff by a materially changed model output. ·