October 2026 so far: every number we published, graded.
The same grading as the record page — our probabilities against the bookmakers' closing line, on exactly the fixtures we priced — for the matches played in October 2026. Lower is better; a coin flip scores 0.693.
The closing line is where the bookmakers' price settled just before kickoff, with their margin removed. It is the hardest public benchmark in football, and we are measured against it rather than against a coin flip or against our own past self.
Two ways to read it. We said is our average probability for the outcome across these matches, and it happened is how often it actually did — if the two sit close, the number means what it says. Score is the strict one: a penalty for every match that is harsher the more confident you were and wrong, ours next to the bookmakers' closing price on the same matches. Lower is better; a coin flip scores 0.693.
Gap is our score minus the close's. Purple means we beat the close on those fixtures; red means the close beat us. A gap under 0.005 is inside the noise of a sample this size and is left uncoloured, because it is not a result.
When we say 45%, does it happen 45% of the time?
Every match gets a home-win probability from us. Group those matches by what we said, then count how often the home side actually won. If our numbers mean what they say, the two columns match.
A difference of a point or three is normal at these sample sizes. Bands with fewer than 20 matches are left out — with a handful of games, a miss says nothing.