Graded record
Wins and losses across every call that reached a final result, expressed with the sample size behind it.
A graded record counts only calls with a settled outcome. Games that were postponed, abandoned or voided are excluded rather than silently scored, and pending calls are not counted until the game finishes.
A percentage without a sample size is not information. Sixty percent from ten games and sixty percent from a thousand games describe very different systems, so both numbers should always appear together.
Short-run swings are normal. Judging a model on a week of results says more about variance than about the model.
Related terms
See it applied
Every published call on AI4Gameday carries a stated confidence and is graded in public once the game is final.