Who would win?
Any two sides, any format, any ground. The Oracle shows its probability — and exactly how each factor built it.
Questions
How does the matchup predictor work?
Every side has an Elo-style rating for each format, learned from every ball-by-ball match in the Cricsheet archive. The gap between two ratings, plus home advantage where it applies, gives a win probability; for T20 the model also weighs recent form and head-to-head, and for Tests it adds a draw probability. The breakdown under the result shows how much each factor moved the call.
Is this the same model as the match predictions?
Yes — the same ratings, weights and draw model behind every match page, validated at 66.6% on 4,220 matches it had never seen. A real fixture can differ slightly once its venue host and the latest results are known; the match page is the official call.
Why can I only compare franchise teams from the same league?
League sides only ever play inside their own competition, so their ratings are measured against each other — an IPL rating and a Big Bash rating are not on a comparable scale.
Why is a team missing?
Only sides with at least ten matches in the format and a match in the last three years get a rating. Teams the archive does not cover cannot be rated.