Beta · research
Beta · research · not a prediction or betting advice · 18+
Model board
The corner and card models' expected totals for matches yesterday, today and tomorrow, Iddaa's main line and the model's probability for that line; once a match ends, the actual total and whether the forecast held appear next to it. Click a row to open the match page.
- Corner model
- — · —
- Card model
- — · —
- Latest data
- —
How to read
"Expected" is the model's forecast of the total corners or card points for the match (card points: yellow 1, red 2); for corners the Home/Away values underneath are the split model's per-team expectation, cards have no team split. The tag on the right, like “9.5 over”, is the model's side at Iddaa's main line; with no pick it says “unclear”. The range (corners: expected ±1; cards: the three most likely numbers) is shown on the match page and its hit rate is measured in the scorecard. The scorecard shows how often it holds.
The line next to it is the main line Iddaa sells for that match; the percentage is the model's probability for over or under that line, with Iddaa's implied probability and price for the same side underneath.
The lineup badge shows the state of the starting XI, and the change column shows how much the card expectation has moved since before the lineup.
The result column fills in when the match ends: actual total corners (home–away next to it) and card points; green ✓ if it landed in the range, ✗ if not. The small line and mark on the right show whether the side (over/under) the model picked at Iddaa's main line held. If the stats have not arrived yet it says "awaiting result". If the model is not sure, the side would contradict the expected number, or it goes against Iddaa, it gives no pick (“unclear”) and that match gets no mark.
Weekly scorecard
Finished matches: the model's last forecast against the actual total. Bias = expected − actual (positive: the model expects too much). Range hit = the actual total landed in the board's range (corners: expected ±1; cards: the three most likely consecutive numbers); past weeks are counted with the same rule. The main-line columns only look at the line Iddaa sold both ways for that match: the model's over probability, Iddaa's implied probability and the actual over rate side by side; lower Brier is better.