A cricket intelligence engine that reads the vitals of the game — player workload strain, clutch-moment impact, momentum, and win probability — built on 1,243 real IPL matches, not surface-level box scores.
Every appearance is time-stamped and matched against the player's rest gap before it. Back-to-back matches (≤1 day rest) are weighted 3× heavier than a normal turnaround, 2-day gaps 1.5×, tapering to zero past 4 days — producing a strain-per-match score that a fitness staff could use to flag rotation candidates before an injury, not after one.
Raw Player-of-the-Match counts reward volume. Our clutch-weighted score triples the value of any POTM earned in an eliminator / knockout stage, surfacing players who perform when the stakes — not just the scoreboard — are highest.
A 15-match rolling POTM rate traces momentum across a career — peaks, plateaus, and declines — the same shape a selection committee would want before a squad announcement.
Conventional wisdom says "win the toss, bowl first" in T20s. The data mostly agrees — but the margin, and whether it holds, changes venue by venue.
| Venue | Toss Call | Win Rate | N |
|---|
Head-to-head history between franchises, all-time.
A logistic regression trained on chronologically leak-free features — each team's rolling win-rate, head-to-head record, and venue bat-first tendency as they stood before that match, never after. Below runs live in your browser using the model's actual learned coefficients.