The 8.5 Threshold: Reading Death Overs Before the Auction Memo Is Written

Two fast bowlers. Death-over economy 8.3 and 8.4. One bought for twenty lakh, the other for two crore forty lakh. In my sheet the gap was not talent. It was definition. One bowler keeps going to the yorker in the 18th over; the other bowls the same over with slower balls and a wide line. Their economy looks identical because economy does not record what the bowler intended. So I keep a second column next to economy: control percentage. The definition is plain. Every death-over delivery is sorted into three buckets — the line and length the bowler chose and executed, the ball that drifted from that intent, and the ball where the batter had already committed to a shot. The first and third count as control. The second does not. Same economy, control of 71 per cent against 52 per cent. Sixty matches, one hostel room, and a spreadsheet no franchise has ever asked for. This piece starts in one column of that sheet and ends on a threshold. Method first, because without it every number below is just opinion. The sample covers 120 matches from the 2026 and 2026 Indian T20 league seasons — sixty home-and-away, sixty in the UAE. The inclusion rule was strict: only matches with ball-by-ball records I could access and watch myself. Every delivery from the 16th to the 20th over was logged on three columns — line, length, and the batter's shot intent. That produced 11,400 deliveries, of which 10 per cent were re-checked. The codebook ran to twelve pages and still sits in the same folder. Limits before anything else. The sample excludes matches I could not attend or watch, excludes deliveries where the camera angle made release point unclear, and carries one flaw I cannot remove: I decided shot intent with my own eye. The data can be reproduced; the same numbers will not emerge from someone else's eye. Conclusions this sample cannot support are not written here. The sheet grew out of the empty-stadium study, which compared 60 UAE matches in 2026 with 60 home-and-away matches in 2026. Home win rate fell from 55 per cent to 47 per cent, and death-overs scoring rose 0.4 runs per over. That study left me with one rule I apply before every death-over claim: a death-over number outside its context means nothing. Where the home pressure is absent, economy improves on its own. Before quoting a death economy I ask which ground, which scoreboard pressure, which batter. Now the threshold. In the 2026 season data I applied two conditions together — economy under 8.5 and control above 60 per cent. Fourteen of 41 Indian seamers qualified, each having bowled at least 24 death-over deliveries that season. Most of the remaining 27 had economy under 8.5 but control under 60 — bowlers whose figures look tidy because the batter erred, not because the bowler did anything. Comparing those fourteen against market averages for contracts and auction payments that year, the gap averaged 1.2 crore rupees. The bowlers who cleared the threshold were paid less than their death-over work was worth. I wrote that memo without being asked and sent it to ten franchises. Ten replied nothing. The memo's real life began when my firm used it in a client briefing and a digital editor saw the deck. Publicly it remains ignored; internally it became part of auction valuation. Six months later the auction answered, and I have no interest in pretending the answer was entirely in my favour. To see why, one rule has to change. The Impact Player has disturbed the economics of the death over. A team used to need two options for the 19th and 20th; now a third sits on the bench and bringing him on costs someone else's place. A death specialist's price is therefore set less by his economy than by how much space he occupies between the eleventh and twelfth options. The threshold alone is no longer enough; it has to be divided by role load. Re-sorted that way, the 2026 data splits into three clean buckets. Low economy, high control, death-only role — the ones the market undervalues and the ones a team cannot buy in place of a fourth overseas slot. Acceptable economy, low control, but capable with the new ball — priced on their first spell rather than their last, which is where the biggest error sits. Low economy, high control, equally effective in the powerplay — priced correctly already, so a memo about them is wasted work. Here is the part that stings. All of this is league data, because league data is public. Ranji Trophy and Vijay Hazare produce thousands of death-over deliveries every season with no public record at all. No threshold can be built there because nobody is counting. A player whose career rests on domestic death bowling has no evidence of his own work. Selection arguments do not stall there because the numbers disagree; they stall because there are no numbers. Now the part where I doubt my own work. The 2026 memo was vindicated, and I will also say it was probably vindicated for the wrong reason. That year the supply of Indian death specialists was thin and several teams needed the same thing at once. Prices rose because demand existed, not because a franchise read my threshold. Correlation and cause sit on either side of that gap, and I will keep repeating it because it is the largest weakness in my method. The second weakness is that my model cannot see a captain's read that a bowler's release point has shifted. In 2026 one bowler sat outside my threshold and was still bought, not for his control but because his coach had spent three months trialling him with the new ball and believed his role was changing. That role change was nowhere in my data. If a coach tells me next season that a release point has moved, I will set the threshold aside and look again. A threshold is a tool, not a verdict. So the next signal. I have already written the failure condition for next season: if the average payment for Indian seamers who clear the threshold reaches or exceeds the market average at the next auction, my memo is dead. On that day I will not claim the market listened. I will say the threshold no longer carries information. And if the gap widens from 1.2 crore to 1.5, the question will not be whether I was right. It will be why ten franchises could not spare one reply.