Final score 1-2. Below is what the model published before kick-off, and how each market was graded.
14 September 2026 at 18:00 UTC
| Market | Model says | Probability |
|---|---|---|
| Match result Miss | Erbil win | 47% |
| Both teams to score Hit | Both score | 64% |
| Over / under Hit | Over 1.5 goals | 84% |
| Double chance Hit | Erbil or Al-Karma | 76% |
| Clean sheet Miss | Erbil clean sheet | 25% |
| Correct score Miss | 1-1 | 11% |
| First goal Miss | Erbil score first | 54% |
| Half time / full time Miss | Erbil / Erbil | 29% |
| Total corners Hit | Over 8.5 corners | 69% |
| Total cards Miss | Under 5.5 cards | 74% |
Each row is a calibrated probability, not a tip. The model is fitted on historical results, re-run when the lineup lands, and every market here is settled and graded once the match is over — the verdict stays on the page whether it went our way or not. Market-level records live on the predictions board.
Every market on every match, the edge against the market price and the AI panel live in the app - and the Accuracy Ledger is free in it.
Get it on Google Play