Who holds their level when the break comes.
Every ball from the sixteenth over on, and every ball of a chase at ten an over. Ranked on one number — PRI — weighted for who they faced and where they played. This is the list we would take into an auction, and the ceiling it is measured against.
The final board, in the order we sign it.
Two top tens, fixed. Names marked GOLD are elite and retained — they are the standard, not the shopping list. Select any row for the full card.
Volume is the defence.
A high rate on forty balls is a rumour. The x axis is how much pressure cricket a player has actually faced; the y axis is what they did with it. The names to the right earned both.
The band, not the point.
Each bar is the 10th–90th percentile of PRI across resamples of the match set. The number beside it is how often the player stays in the top ten when the data is resampled. A wide bar with a low number is the row to argue about in the room, not to hide.
Pressure is measurably different.
A rank is a range, not a fact.
Every published name carries a 10th–90th percentile band from match-cluster resamples, and a retention number: how often the name stays in the top ten when the matches are resampled. A wide band with a low retention is a row to argue about, not to hide.
Read Stubbs honestly. Tristan Stubbs holds his top-ten place in 4% of resamples, with a band of 64.6 to 76.0. He is a low-volatility anchor, not a high-ceiling lock — the board ranks pressure skill, it does not promise he repeats it. The same caution applies to any row whose retention bar is short.
Four roles. Three bids each. One rule.
P1 is the best risk-adjusted name with a 500-ball IPL floor — the bid we can defend. P2 is the next best if P1 is gone or priced away. P3 is the same shape bought below that floor. Risk-adjusted score is the 10th percentile of the bootstrap band, not the point estimate.
Overseas status and availability are front-office inputs; no transfer fee is modelled anywhere in this pack. Quota: eight overseas in a 25-man squad, four in the XI.
Every name, searchable.
The full eligible pool — all names, not just the published twenty. Filter by board, role and slot, sort any column, and export exactly what is on screen. Bands and retention come from the same 400-resample bootstrap as the deck.
Screened out rows are ranked so the screen is visible, not hidden: the IPL does not permit Pakistani players at auction. They carry no bootstrap band because the resampling pool is the signable one — that is a property of the experiment, not a gap in the data.
The ceiling is not for sale.
Read the shape, not the size: he wins by suppressing, not by striking. Every axis is oriented so a longer bar is better.
No bid is modelled for him.
This is an assumption dated to the retention deadline. If he reaches the auction he becomes target one and every table here re-sorts. What we can do is buy the shape of him.
A 90% match on 110 balls is a question, not an answer — which is why the sample is printed beside every name.
The ceiling is the reference. The list is the plan.
The strongest numbers in this dataset belong to players no rival can buy. So this round names the ceiling honestly, then buys the shape of it.
Eleven of the twenty are realistically biddable — and the head of that eleven is not a discount bin: Shashank Singh stands fourth in the entire pool on PRI.
Two piles: sign now, or develop.
IPL exposure is the only buyability signal this pack carries, so it is the split. Above 500 balls buys a proven floor; below buys a ceiling at a discount.
Show the working.
One script, 200,961 deliveries, four corrections, and two caveats we will not hide. Every number on this page regenerates.
PRI = 50 + 12.5·Σwᵢzᵢ
Four things were wrong. Fixed.
Two caveats, both small.
Opposition quality is estimated from the same ball stream — bounded to ±28%, rank correlation with the unadjusted board 0.93 / 0.87. Knockout samples are thin: 4.4% of legal balls, 20 batters clear a 30-ball sample. The knockout column is a tie-breaker, never a ranking input.
Nothing here guarantees a win.
This is a reproducible ranking with published failure modes. What it does guarantee is that every name can be defended with a number, and every number can be regenerated from the raw deliveries. Pakistani players are excluded because the IPL does not permit them at auction — a competition rule, not a judgement about ability.