TipKhel · Guide
How cricket win probability works
What a TipKhel win probability actually means, how the pre-match read is frozen, and how we grade ourselves in public. Predictions, not guarantees.
A win probability is not a promise. If India are given 62% against New Zealand, the model is saying that in 100 similar matches, India would be expected to win about 62 of them — and to lose about 38. That is why we publish a public accuracy record and a model misses page, instead of only the hits.
TipKhel builds a pre-match read from team ratings that move after every result, plus recent form, venue history and (in football) home advantage. If the stronger side is under 56%, or the data is thin, we often publish both probabilities and make no call. Those matches are not counted as hits or misses. The published read is frozen at the toss or kick-off, so it cannot be edited after the match starts.
A format-by-format table will appear here once we have at least 20 graded, non-seed calls in the database.
During a limited-overs match, a live win graph can update over by over. That live line is an estimate only. It never replaces the frozen pre-match read. See DLS explained and how the model works.