Not the Death Overs but Overs 7–15: A Correction from a 63-Match Ledger

In a regular-round league match, a side chasing 176 needed 18 off the last ball of the 19th over. That side's death-overs strike rate finished at 168.0 — 26 runs higher than the team that won. A long-time colleague beside me said the chase was lost because they could not open their arms at the end. The ledger testified otherwise. The problem was not in the last five overs; the match had already been split between overs seven and fifteen, where the losing side played out 54 dot balls and its opponent played out 31. The last-five-over arithmetic explained the result that day, not the cause.

The Khulna ledger did not lie: 63 matches, 2,847 balls, and one quiet conclusion.

In the 2026-18 season I broke down ball-by-ball records from 132 domestic league matches — the BPL, the Dhaka Premier League and the National Cricket League regular rounds combined. I had no software then. On a hand-drawn coordinate grid I plotted line, length and pitch map for every ball, and that grid later became my ball-quality table. Early on I tried to transplant football's xG method into cricket unchanged — in the 2026-18 Dhaka Premier League I produced 1.44 xG per match for Abahani Limited Dhaka's title run against 0.81 conceded, and was rather pleased with myself. Then I understood that the quality of a shot in football and the quality of a delivery in cricket are not the same thing; without a pitch map, xG is only a handsome number.

The BPL began in 2026 with six teams, and from that start the shape of domestic T20 went in one direction — franchise batters, overseas bowlers, a short league phase. The result is that death overs attract the most talk, the most clipped video, and the least record-keeping.

Let me state the limits of the ledger plainly. Of the 132 matches, only 63 saw both innings run to 20 overs; the rest finished early, ended under Duckworth-Lewis, or had incomplete footage. In those 63 matches, 3,412 balls were bowled between overs 16 and 20. Of those, 2,847 balls could be sorted into six execution categories — 83.4 percent. 316 balls stayed unclassified and 249 did not reconcile with the scorecard. I keep the missingness audit beside the data, because forgetting it is how you get caught.

The strike-rate gap between winning and losing sides in the death overs is 6.4. Winning sides' death-overs strike rate was 164.7, losing sides' 158.3. That gap tells you nothing, because every side attacks in the last five overs; the structure of T20 pushes them into it. In the powerplay the picture reverses: winning sides struck at 128.4, losing sides at 131.9. The teams that hit harder in the first six overs lost more often later — in this 63-match ledger, at least.

The gap that actually opens up lives in overs seven to fifteen. Winning sides' middle-overs dot-ball rate was 44.2 percent; losing sides' was 36.8 percent. That seven-point gap is the widest in the ledger — far wider than the 6.4 strike-rate gap in the death overs.

A dot ball does not cost one run; it raises the price of risk in the overs that follow. A 40-run stand over 24 balls and the same 40 runs over 42 balls are separated by more than 18 deliveries. The required-rate slope in the last five overs becomes far steeper in the first case; the batter goes looking for the perfect yorker, plants the front foot early, and that is the easiest way to get out. Fewer dot balls means a gentler climb at the end.

Who bowls those dot balls? The ledger says spin's dot-ball rate in overs 7-15 was 41.8 percent against pace's 35.1. Bowlers of the kind of Shakib Al Hasan or Mehidy Hasan Miraz do work in this phase that looks slow but costs the most in the numbers — because a dot ball in overs 7-15 arrives carrying interest, while a dot ball in the 18th over is only a ball.

Sides that bank a spin over for the last five overs have conceded an average of 11.2 runs in that banked over; the same bowler operating in overs 7 to 15 would have cost 7.4. Spin's dot-ball rate in overs 16-20 falls to 28.4 percent and the economy rises to 10.6. The hand that built the jam through fifteen overs starts handing out gifts in the sixteenth.

Twenty-three of the 63 matches were decided by fewer than 10 runs. A three-and-a-half-run swing in one over could have settled almost every one of those 23. In table language that is four runs in the match equation; in ledger language it is the fate of roughly three regular rounds.

A second thing the ledger surfaced gets less airtime. Bowlers who delivered 12 or more death-overs balls inside a seven-day window saw their economy rise by an average of 1.4 in the next match. There is no need to hunt for a story about sudden loss of form; this is the schedule's arithmetic. When three matches fall in seven days, the rest window allotted to muscle does not survive the calculation. However good the medical staff, lowering schedule load is not their responsibility — it belongs to the calendar.

A caution belongs here, because the cleaner the ledger, the more room it gives you to lie. Dot balls and wins are correlated, not caused. A side with three spinners and a side with one simply have different middle-overs dot-ball capacity — so is dot-ball percentage a proxy for squad investment? Somewhat. The sample is biased too: matches where both innings ran full are the tighter contests; one-sided games often never need 20 overs, so that picture is absent from the ledger. Pitch and venue are playing along — on Mirpur's low bounce the cost of a banked spin over showed up less than in Chattogram, and I have not been able to apply a venue adjustment to every slice.

One more thing needs clearing. There is a story of capability behind banking a spin over. A side with four frontline pacers has no such question to answer. That means the four-run cost can be a tactical error or the structural constraint of a weaker squad — and treating the two as equivalent blunts the analysis.

The error log stays open, because a model without an audit is only an opinion. I wrote three hypotheses down beforehand: that the overs 7-15 dot-ball rate would predict better than powerplay strike rate; that the cost of a banked spin over would be four runs or more; and that death-overs economy would rise in the match after a congested schedule. The third is the thinnest in my ledger, where the sample is just 38 bowlers.

The place to watch next round is therefore not the last over but overs seven to fifteen. Once dot balls start accumulating there, the final column of the scorecard is largely already written — visible in the closing highlights, while the line that got away stays in the ledger. I write the number first, then the argument.