Probability error. Lower is better.
Audit trail
Prediction history that can be checked
We save the first usable public forecast, wait for a verifiable final result, and then calculate trust metrics. This means the number of matches may be small at first.
Early evidence
3 completed matches evaluated
This is a small sample. Treat it as a progress report, not a prediction guarantee. We will call it ready after at least 30 eligible matches.
How close confidence is to actual results. Lower is better.
How often the selected team won.
Supporting penalty for confident misses.
Calibration
Forecast confidence vs results
Forecast bucketAverage forecastActual win rateMatches
0–10%——0
10–20%——0
20–30%——0
30–40%——0
40–50%——0
50–60%——0
60–70%——0
70–80%79.9341%100%1
80–90%85.3881%100%1
90–100%99.6808%100%1
Immutable records
What was actually published
Completed snapshotCricket
JSY vs Nepal
Forecast: 99.68% for Nepal
Result: Nepal
Open immutable record →Completed snapshotCricket
KAK vs RWT
Forecast: 79.93% for KAK
Result: KAK
Open immutable record →Completed snapshotCricket
Gujarat Titans vs Punjab Kings
Forecast: 85.39% for Gujarat Titans
Result: Gujarat Titans
Open immutable record →How to read this
Brier and ECE judge probability quality. Accuracy judges the simple win/loss call. They are different questions, so we show them separately. The ledger uses one first usable forecast per completed match; live ball updates are not counted as independent matches.
Read definitions →