1. Match selection and data
The editorial team selects a match before it starts and reviews the competition, the teams, recent form and the available match context. Core match status and result data are synchronized through API-Football.
- We do not publish predictions retroactively.
- Every prediction is tied to a specific match and the provider’s match ID.
- Form and statistics add context; they never guarantee an outcome.
2. Locked before the match
After publication, KORARAB stores an immutable copy of the prediction. It records the match, original start time, selected market, expected score, odds, editorial confidence and the language versions that were published. A SHA-256 integrity fingerprint helps confirm that this locked KORARAB record has not been silently changed. It is not presented as an independent notarial or blockchain timestamp.
- A new verified copy cannot be created after the match starts.
- Live match data and the final score may update, but the published prediction itself is not rewritten.
- The public verification receipt shows the lock time and the record’s integrity fingerprint.
3. What editorial confidence means
Editorial confidence is KORARAB’s assessment of how convincing the selected match scenario appears. It is not a mathematical probability, a promise of profit or a forecast of future returns. It does not decide whether a prediction is graded as successful.
4. How the result is calculated
A prediction is graded against the market and published pick locked before the match and the regulation-time result. If no separate structured market was selected but the locked published pick maps unambiguously to a supported market, KORARAB uses that immutable pick for automatic grading. By default, this means 90 minutes plus stoppage time; extra time and penalties are excluded unless the market explicitly states otherwise.
- If the condition is met, the prediction is marked “Prediction successful”.
- If it is not met, the prediction is marked “Prediction unsuccessful”.
- If the locked pick cannot be mapped unambiguously to a supported market or result evidence is incomplete, the prediction remains ungraded until review.
5. Postponed, cancelled and unusual matches
The original start time remains part of the publication record. A provider-confirmed reschedule may update the current match time while the locked prediction stays unchanged. Cancelled or abandoned matches are not turned artificially into wins or losses.
6. Statistics without cosmetics
The success rate is calculated only from settled predictions: wins ÷ (wins + losses). Upcoming predictions, matches awaiting a result and records that require review cannot improve the percentage. One match is counted once across all language versions.
- The homepage always shows the sample size.
- Removing a public card does not erase the prediction from the aggregate record.
- Past performance is never presented as a guarantee of future returns.