The Arithmetic of Death Overs: Three Numbers That Tell the Truth Beneath the BPL Regular-Season Table

On the evening of 7 February, sitting in the third row of the press box at Mirpur's Sher-e-Bangla National Cricket Stadium, I entered a number in column 11 of my ledger: 7.42. Four overs, 29 runs, two wickets. On paper it was one of the finest death spells of the regular season. The bowling side lost by six runs. The next morning's reports called it a batting failure.

In my ledger, the location of that defeat is not column 11. It is column 14, where I record pressure-ball percentage — the share of dot balls delivered once the required rate has already crossed nine. In those four overs there were seven dots. Six of them arrived after the contest was effectively settled, when the batting side needed two runs a ball. Column 11 says superb. Column 14 says delayed.

The scorecard counts economy. It does not count pressure.

Context: the columns nobody writes

I joined the sports desk of The Daily Star in 2026 with a notebook and a calculator. In those years a cricket report meant runs, wickets and the smell of a dressing room. Nobody asked how many dot balls fell between overs seven and fifteen.

I stopped writing match reports from memory on a specific day, when I reconciled the numbers and found that a spell I remembered did not match the figures I had filed. Readers had not caught it. I had. From that day the rule was fixed: I do not publish a claim unless a number sits beside it and a sample size sits beside the number.

In the 2026-16 season I volunteered to hand-code an entire season into a single spreadsheet — ball-by-ball state, batter, bowler's line-and-length zone, field placement, and the expected runs generated from that delivery. The ledger surfaced a 23-year-old left-arm seamer whose pressure-ball wicket rate no local scout had ever quantified. He was signed for roughly forty thousand dollars. Eighteen months later he was sold abroad for one hundred and eighty-five thousand. That spreadsheet became my proof of concept and my first paid analytics contract.

I built the first xG chain ledger before the league knew it needed one. In the BPL, the first consequence was dull: I stopped drawing run charts after matches, because a run chart tells you what happened, never why.

In 2026 I published a post-mortem ledger of sixty-four matches within 72 hours of the final. It was not a burial. It was a transfer blueprint. After hand-coding more than seventeen hundred shot events across 33 days, the data showed one side reached the final by conceding 1.4 goals per match below its opponents' expected output. Nobody had written that because nobody had the column. Cricket is where I do exactly this — I write the column the scorecard omits.

At sixty-one, I learned that silence has a crowd coefficient. In 2026, when stadiums shut, I analysed 512 matches across Europe's top five leagues. Home advantage in goals per game collapsed from 0.38 to 0.11; home penalty awards fell nine per cent. When Euro 2026 reopened grounds, I re-ran the model and found the effect returning at roughly sixty per cent capacity. Since then I apply a correction factor to every performance I judge.

This regular season I am running fourteen columns across 46 matches: match number and date, venue, toss result, first-innings score, powerplay strike rate, middle-over dot percentage, death-over economy, wicket-ball percentage, catches dropped, dew index, travel distance and rest gap, and the crowd coefficient. I publish sample sizes, pre-register update rules, and print my misses. This season I logged eleven, six of them on the dew index.

The three numbers

| Metric | Season average | Playoff sides | Gap | |---|---|---|---| | Powerplay strike rate (overs 1-6) | 129.4 | 133.1 | +3.7 | | Middle-over dot percentage (7-15) | 38.6 | 31.2 | -7.4 | | Death-over economy (16-20) | 10.18 | 9.03 | -1.15 | | Pressure-ball percentage | 22.4 | 31.7 | +9.3 | | Catches dropped per match | 1.8 | 1.1 | -0.7 | | Bowling powerplay dot percentage | 44.1 | 51.6 | +7.5 |

The table shows correlation. It proves nothing about cause, and I will come back to that.

Number one: powerplay strike rate is a comfort myth

The gap between playoff sides and the rest in powerplay strike rate is 3.7 runs per hundred balls — effectively nothing, in the most photogenic phase of the innings. I have watched eight matches this season in which a side made 55-plus in the first six overs and still lost. Watching one of them from my home in Barishal, I wrote two numbers side by side: powerplay strike rate 148, and dot percentage from overs seven to fifteen of 49. The second number had already called the result.

Powerplay strike rate does not separate teams because a powerplay is only six overs, while the middle phase is nine. The largest block of the innings is the least measured.

Number two: dot percentage from overs seven to fifteen

Playoff sides ran a middle-over dot percentage of 31.2. Everyone else ran 38.6. That is 7.4 points, roughly seven extra dot balls across nine overs. With death-over economy currently at 10.18, seven dots left unaddressed can cost eleven to fourteen runs later. In a T20, that is usually the match.

Who generates those dots? Four of my five best middle-over spells this season belong to spinners, three of them left-arm. The reason is structural: from overs seven to fifteen, the slog-square and long-on fielders can be pulled in, because powerplay fear is gone but sprint fear has not begun. The side that turns that gap into a coaching plan sits high on the table.

Number three: pressure-ball percentage

I built this column because death-over economy is an average, and averages do not understand pressure. Pressure-ball percentage measures how often a bowler produced a wicket or a dot once the required rate crossed nine. The season average is 22.4; playoff sides ran 31.7.

I follow the ball before the shot, because the chain explains the finish. The 7.42 economy from my opening paragraph is only valuable if column 14 confirms those balls came at the right moment. A bowler who concedes nine in the sixteenth over and four in the nineteenth has the same average and a different function.

Catches dropped matter too, and lightly dismissing them is a mistake. Playoff sides dropped 1.1 per match; others dropped 1.8. Across seven matches that is roughly five extra lives, and in T20 five lives often decide a tournament.

Dew, travel and crowd: the correctors

No performance can be judged raw, because BPL venues are not equal. Mirpur gives dew in the second innings at night. Chittagong's wind bends right to left, easing reverse swing for left-arm quicks. Sylhet's faster outfield rewards cover running. Columns 12, 13 and 14 exist for exactly these three things.

This season, sides batting second at Mirpur scored on average 7.8 more in the first powerplay, and that shaped toss decisions. But it was not true everywhere: the effect appeared in twenty-nine of 46 matches. Not thirty. Here I testify against myself — extend the rule beyond twenty-nine and it stops being analysis and becomes superstition.

Crowds returned this season, so the crowd coefficient is live again. Home win rates came in above away rates, but the gap is smaller than in football. In cricket the crowd works mostly through umpiring and through bowlers' over-rate psychology, not directly through runs. The crowd coefficient taught me that absence can be measured as loudly as presence, which is why closed-door matches remain a separate sample in my ledger and are never deleted.

The transfer ledger: the arithmetic of regret and opportunity

I do not manage transfers; I manage the arithmetic of regret and opportunity. Every transfer rumour enters my ledger as a probability, not a promise. This season I price players in three layers: broad outcome (strike-rate splits by batting position and innings number), relative impact (contribution above or below the match run rate), and projection (dew, venue, and opposition bowling zones).

One domestic spinner in my ledger was not defined by his four-wicket haul but by the dot balls he manufactured in overs seven to fifteen — an average of eight dots a four-over spell, five of them delivered when his side was already under pressure above sixteen. The auction priced him off his death-over economy. That, to me, is the wrong column. He built the pressure in overs seven to fifteen; the auction paid for the delayed phase. What is easy to demonstrate gets the money.

Contrarian: correlation is not causation

The alternative explanation is entirely possible — good teams lead, so their spinners can bowl dots; bowling dots does not make a team good. That is the primary worry. Forty-six matches is a small sample, and twenty more could move it. Second, spurious links: sides that played more pressure matches also bowled more dots, because exposure produced the number, not quality. Third, conditions are not equal across six franchises, so I have tracked which sides avoided dew-affected matches. The toss is a large cause, and the toss cannot be pre-read.

I do not hide my misses. Pricing death-over economy this season, I correctly identified four of six contenders and got others wrong. A post-mortem ledger is a confession written by the data after the final whistle — it records what happened, what could have happened, and what never did.

What to watch

One question for the next phase: will middle-over dot percentage become the auction's primary currency? If it does not, the BPL market is still not measuring what wins. I cannot say whether prices will correct. I can say my column 14 is ready for next season, and it will make fewer mistakes than the last three. The market sprints. The ledger waits.