The two top-10 rankings from my Round 1 analysis: the batters and bowlers who get better when the game is on the line, measured ball by ball against their own usual level.
Ordered by the model score. Tap a player for the full pressure record.
Delivery rate: share of pressure matches at or above the player's own usual strike rate.
Every one of the ten scores faster when the pressure is on. Part of each uplift is the mechanical death-overs effect (17% pool-wide); adjusted for that phase effect, Stubbs still beats his phase-matched expectation by about 44%. Raza, Buttler and Dhir delivered in every qualifying pressure match.
Uplift is the raw ratio of pressure strike rate to usual strike rate; the model scores the shrinkage-adjusted ratio, which is a few points lower (Stubbs +50% after shrinkage).
Ordered by the model score. Tap a player for the full pressure record.
Delivery rate: share of pressure matches at or above the player's own usual economy level.
The raw model ranks Tim Southee second, but his record in this pack ends in February 2024 with no 2025 or 2026 data, so he was removed on review. The bowlers below him move up one place and Noor Ahmad, whose record runs to November 2026, takes the tenth spot. This is a deliberate recency-informed selection decision, not a change to the model itself.
Death overs cost more for everyone - 97% of eligible bowlers concede more under pressure than usual - so bowlers are judged on deteriorating less than the pool, then on wickets and repeat delivery. Theekshana holds the best pressure economy in the eligible pool while still improving on his own baseline.
Flag pressure deliveries from the match situation: required run rate, wickets in hand, balls remaining and stage of the innings. 29.5% of batting deliveries and 19.8% of bowling deliveries qualify.
Measure each player on those deliveries against their own usual level, so a player who faces tougher attacks is not penalised.
Shrink small samples toward the pool, then score uplift, absolute performance, match-by-match delivery rate and volume together.
Require enough pressure exposure that one hot series cannot carry a ranking: 180+ pressure balls, 8+ pressure matches, and at least 5 (batters) or 8 (bowlers) matches with enough balls for the match-by-match delivery rate.
The whole method was re-run on IPL matches only and on 2024-onwards matches only. Where a pick drops out of a stress-test list (Kuldeep, Ngidi, Russell, Rauf and Noor Ahmad on recency, six of the ten bowlers on IPL-only), the written case says so openly.
Naman Dhir (210 pressure balls) and Sikandar Raza (304) have flawless but thin records; Lungi Ngidi delivers at his usual level but inconsistently match to match; Andre Russell ranks as a part-time bowler; Noor Ahmad's record is current but smaller than the specialists above him. These are the places a judge should push, and the report flags each one.
Reserves, in order: Tim David, Marcus Stoinis and Heinrich Klaasen (batters); Mitchell Starc, Shaheen Afridi and Harshit Rana (bowlers).
These are performance rankings from historical IPL and T20I data for the Round 1 task, not an auction-availability shortlist.