Auctions Price the Last Four Overs; the Truth Lives in Overs 7–15

On 19 December 2026, in Dubai, the paddle was climbing on the auction floor while I sat in a data room in Bangalore, opening my phase-split sheet. When Mitchell Starc's price settled at INR 24.75 crore, it became the highest bid ever recorded at an IPL auction, and Kolkata Knight Riders had bought him. For anyone watching the feed at home, the story was simple: a World Cup-winning pacer, experienced, a man for the big night. My screen asked a different question: which balls actually produced that price? If I put Starc's league-phase spells and those three playoff spells in the same box, the number does not survive. It survives only out of the second box. The clips sitting in the market's hands were playoff clips, not league clips. The auction did not buy a cricketer; it bought a memory — and memory is the dataset with the smallest sample size.

Before reading this window, the structure needs to be laid out cleanly. In the IPL cycle, a player's price is set in three stages: retention, trade, auction. Retention prices him through the franchise's internal valuation, trade prices him through the collision of two teams' needs, and the auction prices him publicly, when several teams bid on the same player at the same moment. That is where the strange thing happens: the information that gets the most airtime gets the most money — however good the information is, airtime sets the price.

Salary caps, overseas quotas, NOCs and availability windows decide who can be bought. What decides the price is the broadcast schedule. The last four overs sit in prime time, finals night trends, and the seven middle overs play out under a neutral camera angle, in the shadow of an ad break. The market lifts its price exactly where it points its lights.

My own habit has been the same since 2026. In that Bengaluru FC season I built a live xG and PPDA dashboard, and by matchday five I could see that Sunil Chhetri's goals sat on relatively low xG while Miku's sat on higher xG. One was scoring more from fewer chances, the other scoring less from more. I wrote about Miku's looming regression that season, and it arrived. I carry that loyalty to a model into cricket carefully — ball-by-ball data, innings structure, format and wicket equity give what football xG does not. Ignore that difference and analysis turns into storytelling.

I divide a T20 innings into three blocks: powerplay (1–6), middle (7–15) and death (16–20). The question is which block moves the result most. In my phase model — the basis of the pressure index I use on air — the largest win-probability swing historically occurs between overs 7 and 15, especially across overs 10 to 15, when wickets fall and the run rate either comes under control or slips away. Death overs give the audience drama; middle overs decide the batting order's fate.

This is where the arithmetic of sample size matters. A death specialist might bowl thirty-six to forty-eight balls in that phase across an entire season — eight to twelve overs. Concede fifteen in one over and the economy of that spell jumps by one and a half to two runs. A death economy is a conclusion drawn from about forty balls, and the auction puts eight crore or more on its shoulders. Statistically this is not an excuse; it is a design flaw. The dashboard was never a prophecy; it was a confession booth, where the market's volatility is admitted in public.

The second idea is wicket equity. The wicket of a set number three in the twelfth over and the wicket of a number nine in the nineteenth are not the same asset. The scoreboard shows both as a 'W', and the auction tracker counts both in the same column. In my model, a middle-overs breakthrough adds far more win probability than a death-overs dot ball, because it suppresses the run rate and leaves usable balls for the batter who follows.

The third idea is control. I do not use the word 'pressure' unless I can measure it. Nobody 'owned' the middle overs; they were audited in real time, and like most audits, the report went largely unread.

Take Sam Curran. At the auction held in Kochi on 23 December 2026, Punjab Kings paid INR 18.5 crore, and that price rested on the sum of two roles — the new-ball attack and lower-order batting. The all-rounder premium is real, but it is an average; price the average and you buy two different futures at once, and if one role fades, half the fee is already in the air.

The asset the market prices lowest is the bowler who operates between overs 7 and 15 — the spinner or the cutter — the phase where boundary probability is counted most carefully and the camera visits least. In my count, that bowler generates roughly one and a half times more pressure events across a season than a death specialist, and is paid forty to sixty percent less. This is not hidden skill; it is unequal attention.

The same logic runs through batting. The number three or four who rotates strike against spin through the middle overs, holding the run rate in place, produces no highlight — no six, no clip. Yet he lays the innings' foundation. The batter who makes eighteen to twenty-two off the last ten balls is paid more; the one who makes fifty off fifty between overs 7 and 15 is paid less. Which of the two wins more matches is clear in my model.

A word on method. My pressure index stands on three layers: dot-ball ratio, breakthrough probability, and yorker accuracy in the slog overs. Context variables include pitch condition, dew, limited overseas availability, and the opposition's batting depth. If my model cannot defend a claim, I do not publish it — that single rule has killed a lot of elegant sentences in my drafts.

The fourth layer is economics. For overseas players, NOCs, series calendars and the priorities of national boards decide who is available for how many matches. A franchise has to do the division: if a star available for eight matches costs the same as a player available for twelve, the per-match cost is far higher. There is another reality — for structural reasons, certain players are almost absent from certain markets, and that absence is never a verdict on talent; it is the product of contracts and political economy. Born in Pakistan and working in India, I see this from both sides: franchise valuation happens player by player, not as a team ledger. A coach picking a squad under quota pressure, or under job-security pressure, buys names rather than leverage.

One thing a salary cap cannot measure is bidding momentum. When two teams need exactly the same role, the price stops tracking the player's quality and starts tracking the scarcity of that role inside the auction pool. The same bowler goes for four crore in one auction and twelve in the next — the difference is not his skill, it is how many like him sit on the list. That confusion is the market's most expensive habit.

The Impact Player rule has muddied this further. Teams are no longer obliged to carry a sixth bowling option, so they buy single-role specialists. The consequence is that the middle overs get a shrinking share of bowling allocation, because four overs must be delivered regardless, and the best four are reserved for the phase where the cameras live. It is a clean example of structural incentives destroying a player's value.

So does the death specialist have no value? He may — and here I stay careful. The repeatable skills are release-point consistency, yorker accuracy under fatigue, and matchup history against specific batters. In my count these three variables hold up across years, while their weight in auction pricing is close to zero. Video shows outcomes, the market sells outcomes, and value should sit with process. A spell's outcome and a bowler's process are not one thing; when two quantities move together, that is a correlation, not a cause.

Let me state confidence levels plainly. On predicting auction prices from strike rate and economy, my confidence is moderate, because the sample is small and selection bias is heavy. I keep the alternative explanation open too — perhaps death specialists do possess a narrow but genuine skill that my sample fails to detect. But move the same player from one season to the next, into a different team, a different role, different ball conditions, and the economy often shifts by thirty to fifty percent — there my confidence is high, because the sample is larger and the method simpler.

What I do not write matters as well. 'Finisher premium' and 'death specialist' are, in most cases, artefacts of survivorship bias. The men dropped on bad days leave no data in the market; the one who won a final keeps his clip in circulation forever. To cut that bias I apply a single condition: calculate with showcase innings removed. If a player's price survives only on the last four overs and the playoffs, that is not a price — that is beta risk.

One more misconception needs breaking. Franchise team selection is never a purely cricketing decision. If a coach's job depends on season results, he buys the player who delivers fastest — the one with eight wickets in eight matches, even if another stayed consistent across three. Under that incentive structure, long-horizon valuation takes a hit, and the price is paid by the smaller-budget sides.

In the next window my eye will be on one place only: the contracts of overs 7-to-15 bowlers, and the retention value of number three and four batters. If those two categories rise in price, the market is learning. If they do not, the market's visual bias survives another cycle. Who makes the highlight reel is not cricket's question; who stays outside the highlight reel is the real one.