Accuracy

One test matters: when the model says something is 60% likely, does it happen about six times in ten? Every prediction in a graded market is graded on the final whistle and nothing is edited afterwards. Corners, cards, fouls and half-time markets can't be graded from the final score, so they are not in this record.
Model saidPredictionsActually happenedDifference
90–100% likely19987%−5.9
80–90% likely109882%−2.3
70–80% likely274374%−0.8
60–70% likely227662%−2.7
50–60% likely264256%+1.5
40–50% likely338546%+0.9
30–40% likely569138%+3.5
115098
Predictions graded. A price can be updated until kick-off; the one graded is the last price before kick-off.
±2.2
Average gap between what the model said and what happened, across 18,034 predictions at 30% or likelier, weighted by band size.
0
Numbers changed, hidden or backdated after a result.