The Khulna Ledger: 46 Matches, 11,234 Deliveries and the Polite Lie of Death-Over Economy

March 12, evening, the press gallery at the Sher-e-Bangla National Cricket Stadium, fourth row. The 18th over of the 41st match is closing and I am re-checking a number in my notebook. Khulna Tigers' death-over economy reads 8.94 — the lowest in the league. On the table: six wins from twelve.

On the same notebook, on the same evening, another line had to be written: teams conceding above 10.40 an over between overs 16 and 20 won, on average, two matches more. What a bowling coach calls a certificate of success is, in the table's own language, nearly inert.

One over's number is not a trend. So before reaching a conclusion I sat down again with the full ball-by-ball ledger of 46 matches. The Khulna ledger did not lie: 46 matches, 11,234 deliveries, and one quiet conclusion.

Context: how the ledger was written, and where it leaks

I hand-coded every ball of all 46 matches — 11,234 deliveries in total. Five columns per ball: over number, bowler type (spin or seam), batter's hand, pitch state (new, used, slow) and outcome — runs, dot, boundary, wicket, extra.

I split each innings into three phases: powerplay (overs 1–6), middle (7–15), death (16–20). The two core units are runs per over in each phase, and the price of a wicket in each phase, meaning how many runs were spent to buy one. I computed Pearson correlation against win-loss and, because the sample is small, logged the confidence interval next to every coefficient. A coefficient never stands alone; its interval stands with it.

It matters to publish the gaps. Of 46 matches, three had incomplete event feeds — two rain-affected, one with a broadcast feed failure. For those three I took powerplay strike rate from manual scorecards but discarded the ball-by-ball death sequence and window-based bowler rotation entirely. My death-over sample, then, is effectively 43 matches, not 46. The slide with the largest number on it deserves the most suspicion.

The second gap is dew. After dusk on a used Mirpur surface, dew strips grip from the seamers, makes the ball skid on for spinners, and makes batting simpler in the second innings. I noted an approximate dew arrival over in each match, but nobody measures it officially. So toss, light and dew entered my calculation through the door of estimation, not the door of sensors.

The third reality is where data lives. The digital outlet that first published my Khulna ledger shut down entirely in July 2026. Since then I keep my own copy of every dataset, on multiple drives, in date-stamped files. Platforms disappear; a ledger does not survive unless you keep it yourself.

A model that runs without an audit trail is not a model — it is an opinion. So this piece ends by stating where my own numbers may fail.

Core: where the match is actually written

Powerplay is the indicator; death-over economy is decoration.

The clearest line in the ledger is in the first six overs. Over 46 matches, the correlation between powerplay run rate and winning is 0.61. In the same league, death-over economy correlates with winning at 0.18 — statistically almost inert. Middle-over economy lands at 0.29, a better indicator than the death overs but nowhere near the powerplay.

Khulna Tigers' powerplay run rate was 7.21 against a league average of 8.09. They were roughly one run per over behind before the fielding restrictions lifted, and that gap never came back, because a side already behind in the powerplay is forced into spin-risk in the middle overs, and risk costs wickets. In my ledger, sides scoring under 50 in the powerplay won 23 percent of matches; sides above 55 won 61 percent.

The league champion's powerplay run rate was 8.63, with a boundary share of 21.4 percent. Khulna's boundary share was 14.8 percent. The difference between the two sides' death-over economy was 0.3. The gap was created at the very start of the innings, not at the end.

The price of a wicket in the middle overs

Overs 7 to 15 — nine overs people rarely remember, and exactly where the most political word in T20 hides: the dot ball.

Across the league, a middle-over wicket cost 23.4 runs on average. The four sides taking more than two wickets per innings between overs 7 and 15 won 68 percent of their matches. The four taking one or fewer won 31 percent. That is a 37-point spread — roughly six times larger than the spread produced by death-over economy.

Khulna's middle-over dot-ball rate was 38.2 percent, second highest in the league. That number sounds healthy, but its correlation with winning is only 0.22. Unpack it and the reason shows: their dots came on slow surfaces, on safe lengths, with the scoreboard under no pressure. A side that bowls dots without taking wickets is slowing the game and buying time for nobody — in T20, time belongs to no one.

Correlation between wicket price in the middle overs and winning: 0.54. Between dot-ball rate and winning: 0.22. Placed side by side, these two numbers dismantle an old assumption: pressure is created by dot balls, but matches are changed by wickets.

The polite lie of death-over economy

Now the real question. Why does a low death-over economy fail to convert into wins?

First, game-state padding. In my ledger, sides that were 70 runs or more behind after 15 overs conceded 8.38 an over between overs 16 and 20. Sides still in the contest inside 70 conceded 9.91. When the game is effectively done, the bowling looks good — because the batter carries no risk, because his team has already won. These dead overs are what makes a death-over economy look beautiful.

Khulna's death economy in their six wins was 9.31; in their six defeats it was 8.62. Their economy in losses was better than in wins. In one of those defeats they began the 19th over 22 runs behind — spreading the field and saving a single run is easy then, but the match does not come back.

Second, the accounting of death-over wickets is different. In my ledger, each wicket between overs 16 and 20 cost 11.7 runs — cheap, because it arrives when a batter is throwing his hands. A middle-over wicket costs 23.4. Two death-over wickets are worth one middle-over wicket. That single sentence is the centre of my whole ledger.

103 dropped catches, unevenly distributed

Across 46 matches the ledger records 103 catches dropped. The list is lopsided. Sixty-one of them fell in the powerplay or middle overs — where a wicket is most expensive, fielding was most porous. The forty dropped in the death overs mattered comparatively less, because by then the ball travels off the edge and is catchable in the outfield.

Khulna Tigers dropped 14 catches in the middle overs, the most in the league. I have no tracking data, but my notes show their slip-and-point cordon changing almost every match — no settled catching pair ever formed. Without a fielding plan, drops are not accidents, they are a pattern.

Spin versus seam: the pitch divides the table

On used Mirpur surfaces, spinners conceded 7.1 an over and seamers 8.6. But a seamer's wicket cost 14.2 runs, a spinner's 19.8. Seamers leak more and take more — and in the middle overs the wicket is worth more.

Even so, spin's share of middle-phase overs across the league was only 38 percent. The fear is understandable: a left-hander can take a middle-over spinner apart and swing the match. But my ledger says sides bowling more than four middle overs of spin on used pitches took 1.3 more wickets on average while conceding only 0.7 more runs per over. The risk is reputational; the reward is on the table.

Why I read this differently: correlation is not causation

Now the most important part — the one people skip.

Forty-six matches is my sample. At that size, a coefficient of 0.18 means only this: the tendency of death-over economy and victory to move together is weak. It does not mean good death bowling is worthless. In a final or a semifinal, game-state padding disappears, and every run saved in overs 16 to 20 becomes genuinely expensive. League phase and knockout phase are two different numbers from one household, and most people sit them on the same line.

The second risk is squad turnover. Two sides changed overseas bowlers mid-league and one spinner picked up an injury. The powerplay relationship I found therefore contains squad churn inside it. In today's registration window a side can field two different bowling combinations in two matches, and that change never appears in my ledger as a match number — only as balls.

The registration window has another effect that rarely shows in run rate but does show in over splits. Overseas pace bowlers added mid-league arrive with less than three weeks of preparation; in their first two matches their death-over economy is roughly 1.9 worse. That error belongs to the calendar, not the team. Nobody logs over-load management, because it does not show in one scorecard — it shows three matches later.

And the final warning: records live in matches, tendencies live in seasons. Forty-six matches can analyse a league; they cannot declare a truth. My coefficients will move next season, and that is normal. What will not move is the method: count the balls, split the phases, price the wickets, and publish the limits of the sample. Russia 2026's error log, the 2,412 matches in empty stadiums — those are transfers of method, not of subject; a cricket ball-by-ball ledger stands on the same footing.

Takeaway: what to watch in the next round

Over the next three weeks, look at powerplay strike rate, not the table. A side that cannot reach 50 in the first six overs is carrying relegation-level risk no matter how pretty its death-over economy looks.

The second indicator: wickets per match between overs 7 and 15. One, two, zero — those three numbers will tell you who is actually holding the game. The data does not say who will win; the data says who is no longer in the match, and that is what the rest of the season will be watched for.

And what happens first: death-over economy will be sorted into a neat blue table, someone will photograph it and post it, and the next day the neatly sorted side will lose. The ledger will be open again that day.