Does a 60% favourite win six times in ten? We checked our own numbers
Before every match we publish three numbers: the chance of a home win, a draw and an away win. If those numbers mean anything, a team we give 60% should win about six times in ten. We went back over the 2,834 matches we called before kick-off from 19 July to 25 September. In the middle of the range they hold up well. At the top they do not: teams we made clear favourites, at 70% or more, won 62.0% of the time, not the 78.2% we said.
Said against happened
| We said | Average | It happened | Calls |
|---|---|---|---|
| 0–10% | 6.5% | 17.4% | 138 |
| 10–20% | 15.7% | 22.5% | 681 |
| 20–30% | 25.7% | 25.3% | 3,255 |
| 30–40% | 33.6% | 33.3% | 2,170 |
| 40–50% | 43.8% | 44.8% | 1,307 |
| 50–60% | 53.7% | 51.9% | 524 |
| 60–70% | 63.6% | 56.7% | 240 |
| 70–80% | 74.2% | 60.0% | 120 |
| 80–100% | 85.6% | 65.7% | 67 |
Where the numbers hold
Most of what we publish sits between 20% and 60%, and there the model is close to honest. Favourites we gave 40–60% (1,803 matches) won 47.2% of the time, against the 46.6% we said on average. Across the 20–60% bands the widest gap between said and happened is 1.8 points, inside the margin those bands carry. That includes the draw, which models often get wrong: across all 2,834 matches we gave it 25.7% on average, and 25.9% of them were drawn.
Where we are too sure
Above 60% the dots fall below the line, and the further up, the further they fall.
- Favourites at 60% or more: 427 matches, we said 70.0% on average, they won 59.0%.
- At 70% or more: 187 matches, we said 78.2%, they won 62.0%.
- Big favourites (60% or more) playing at home won 61.7% (350 matches). Big favourites playing away won only 46.8% (77 matches), although we gave them 67.3%.
- In those matches the draw came up 22.5% of the time; we had given it 17.0%.
The same thing shows at the other end. Outcomes we gave less than 10% happened 17.4% of the time, against the 6.5% we said. Put simply, our biggest mismatches are less of a mismatch on the pitch than on paper.
Not one competition
The obvious suspect is the summer qualifiers. The Conference League gave us more big favourites than any other competition, 77 of them, all in its qualifying rounds, and they won 59.7% against the 70.8% we said. But take them out and the rest fall short by the same margin: 350 big favourites from every other competition (by coincidence, the same count as the home favourites above), 69.8% said, 58.9% won. The gap is in how the model turns a big difference in strength into a probability, not in one corner of the calendar.
The clearest place it shows is away from home. Of the 77 big away favourites, fewer than half won. Until 22 September the model used one home advantage for every league; that is one of the things the update changed.
How sure are these numbers?
The bands in the middle hold thousands of calls, so their margin is a point or two. The top bands are smaller: the 70–80% band has 120 calls and a margin of about ±8.8 points. Even at the edge of that margin it stays below what we said, so this is not chance. How far below is the part to read loosely.
What changes
Nothing here is hidden after the fact: every one of these predictions was logged before kick-off, and the track record lists them. On 22 September we updated the model: it now uses each league's own scoring rate and home advantage instead of one figure for all, and anchors European clubs to their UEFA coefficients. Only 24 of the matches above were called with the new version, far too few to say whether it helps; we will run this check again once it has a few hundred. Until then, read our biggest numbers as a little lower than they look, and read a big away favourite as a lot less certain. The methodology sets out how the numbers are made.
How this was counted
Every match our model called before kick-off between 19 July and 25 September 2026 and that has a final score, across all the competitions we cover. Results at 90 minutes. Each match gives three probabilities, one per outcome, so the chart counts 8,502 calls. A favourite is the side with the higher of the two win chances. The margins are 95% intervals.