The Silence of Khulna: What the Unwritten NCL Archive Refuses to Record

I was sitting in the western gallery of the Sheikh Abu Naser Stadium in Khulna one November morning, a half-charged laptop on my knees and a glass of tea beside me. A left-arm spinner from Khulna Division took seven wickets in the first innings. In the next morning's paper that became two lines of deserved praise. But on the handwritten sheet kept with the scorecard was something nobody printed: twelve consecutive overs in the second innings had no ball-by-ball entry at all. Only runs, wickets, and a small cross. Which delivery turned, which one the left-hander swept, which over the field changed — no answers.

Those twelve overs are worth more to me than the seven wickets. Nearly every "truth" that gets told about cricket exists because someone wrote it down ball by ball. In Bangladesh's domestic game, that writing does not happen. Since I left staff employment I have sat in whatever ground I could reach each NCL season, and for the ones I could not reach I have coded ball by ball from photographs sent by local scorers. The numbers were not lying; they were waiting for a better question.

The archive nobody reads

The National Cricket League began in the 2026-01 season, at almost exactly the moment Bangladesh gained Test status. Eight teams — Dhaka, Dhaka Metropolis, Chattogram, Khulna, Rajshahi, Barishal, Sylhet, Rangpur. Four-day matches, six or seven per side across three rounds. A full season bowls roughly seven to nine thousand overs.

How much of that becomes a record?

The BCB's own scorecards exist for almost every match — runs, wickets, overs, batting order. That is necessary and not sufficient. To know which bowler bowled how many overs at which batter, who actually turned the ball, at which stage of an innings the scoring rate broke — you need ball-by-ball logs. My own archive holds 41 matches across four seasons, all hand-coded. Forty-one matches is less than a quarter of what four seasons contain.

This is the real data problem in Bangladeshi cricket. The shortage is not of information; it is of preservation. Information vanishes at the instant it is created. Nowhere else in the subcontinent is this so acute. Much of India's Ranji Trophy is streamed, stump cameras exist, Kerala and Mumbai domestic cricket gets podcasts. Pakistan's Quaid-e-Azam Trophy scorecards are digital. Sri Lanka's domestic league is on camera. In Bangladesh a first-class match in Khulna or Barishal still starts in front of three hundred people, and no camera goes there.

I am not here to complain. I am here to say: because the camera does not go, whoever codes it by hand first will see something new. In Khulna I learned that silence is also a dataset.

Spell length: the curve nobody draws

The first thing that collapsed in my log was something I had not expected.

I assumed young domestic fast bowlers were given short spells — four overs, six overs, kept on a leash. After the coding, the picture was the opposite, and unevenly so: the distribution is bimodal. A large share of bowlers under twenty-two are asked for eight to twelve over spells, while their exact age-mates at the next ground never bowl more than four in an innings.

The difference is not talent. It is the depth of the bowling attack. Where three senior seamers are fit, the young one is rationed. Where two are injured, the twenty-one-year-old is the one who has to bowl sixteen straight, whether or not he is prepared for it.

I counted the over-load of eleven young seamers across their first two seasons. Among those who were heavily used — more than ten overs per match in their debut season — the incidence of serious injury over the next two years was noticeably higher. The sample is small, and I will not claim causation. But the association is clear enough that the question cannot be skipped.

The question is this: do we even know who is bowling too much?

We do not. Knowing that requires writing down every spell, and nobody is writing. We hear from television when Nahid Rana's speed crosses 140. Nobody writes about the left-arm seamer in Khulna who bowled nearly a third of his side's overs across nine consecutive matches last season because the rest of the attack was injured. His name is not on a franchise trial list, and his over count does not exist.

If nobody counts a bowler's load, the load does not decrease — the injury simply stops appearing as data.

The peak curve: the age we imported

Here is the most uncomfortable part of this piece.

International cricket runs on a borrowed rule about peak ages — 24 to 27 for batters, 25 to 29 for fast bowlers. Those numbers come largely from English, Australian and South African first-class data, where tempo, annual match volume, injury management and pitch behaviour are all different.

In my NCL log, the peak sits elsewhere.

Across four seasons of innings-level rates, domestic spinners reach their best sustained returns between twenty-six and thirty-one, sometimes beyond. The reason is guessable: spin is a game of anticipation and patience, and on Bangladeshi pitches it rests on weight and rhythm accumulated over years, not finger strength. Fast bowlers, inversely, lose their effective window earlier than the imported model suggests — the work demanded of them, low bounce and heavy effort ball after ball, is paid for quickly and depreciates quickly.

Now ask the question. If a Bangladeshi spinner's best years are twenty-six to thirty-one, and selectors believe a curve in which twenty-nine means the final phase, what are we actually doing?

We are trimming our best bowlers immediately before their window opens. Read Mehidy Hasan Miraz's career as a counter-example: he was established very young and then held his rhythm for a long stretch, because someone gave him time. Our system's impatience is exactly why the curve matters. A peak curve is not human biology; it is a socio-economic contract — where the league offers security, the peak arrives later, and where it does not, a twenty-five-year-old is already labelled old.

The selection window: a cohort, not a generation

In the middle of the last decade we began telling ourselves a very comfortable story. The golden generation.

In my log, it is a cohort. The difference is not merely vocabulary.

Look at the birth years. Mushfiqur Rahim, Shakib Al Hasan, Tamim Iqbal, Mahmudullah — born within a few years of each other, and debuting internationally within a few years of each other. That is not a miraculous flowering of generational talent; it is the output of a selection window. In the first decade after the NCL began, one gate opened: a clutch of young players was poured into first-class cricket at the same time, under the same pressure. Those who fell inside that window filled the national team. When did the next window open? How many came through?

This is the area I test hardest, because it is where almost everyone gets it wrong. Starting from the golden-generation explanation produces the question "why were those five so good?" The correct question is "why did the window after theirs close?"

The answer is probably selection inertia. Once a decade occupies the national side, the domestic league can no longer keep that space open for its own young players. A twenty-four-year-old batter sitting in the NCL just sits, because four men are ahead and all four are playing important matches. That is not individual failure. It is a structural constraint of a pipeline. When Soumya Sarkar or Anamul Haque were in their best years, the question was never about them — it was about the four men behind them, and how many overs they got.

The session that never became a score

Some of my most valuable entries are negative.

Take a match in 2026. Khulna needed three wickets with three sessions left. Rain arrived in the second session of the afternoon. The match was drawn without another ball. On the scorecard it is one line: Match drawn. The bowler's account ended without the game happening; his innings load never entered a season table, his spell length never joined a record. In reality, the pressure that had accumulated in his chest stayed exactly where it was — only the data is missing.

This is the cruellest property of data: absence is never recorded as zero, it is recorded as nothing at all. Sessions lost to rain, matches lost to injury, days suspended for security — all invisible in our analysis, and that invisibility changes the character of the whole sample. The more innings are truncated, the more a "completed-innings average" is an artefact. So at the foot of many new analytics sheets I write one line: how many overs is this, and how much of those overs have we never seen? The real question is never "what is the best average?" but "what fraction of the denominator is visible?"

Correlation is not cause: the question I put to my own dataset

As a working habit, before running any hypothesis I write the prediction and the expected result in advance, so that I cannot rewrite the story after the fact.

At the end of one round I found a spike: domestic spinners' suppression of scoring rate looked better than the same stage of the previous season. I nearly wrote it up — "domestic spinners are adjusting to conditions."

Split by venue, two-thirds of the improvement came from a single ground, and only three matches were played there across the whole round. Three matches. Three matches is a sample for building an argument, but three matches is not a sample — three matches is a coincidence. I did not publish it, because publishing would have made my model bigger than my work. The spike got spiked, but the pattern stayed in the data. It is real, and it is much smaller than the spike suggested.

This is where the biggest error about "home spin dominance" hides. The claim is: spin dominance at home tells you nothing. What tells you something is how many matches were played at a ground, and who bowled there, how much, and in what state.

I apply this rule to myself. I once opened an English draft with a line that still sits in my notebook: you can read a cricketer's role off a heatmap. After hand-coding a few NCL innings I understood that a heatmap is a picture of where balls happened to be, not a picture of intent. The batter who saves two or three extra balls per innings and the one who protects the under-edge and adds runs both glow identically. Role is legible from the spin map, not the brightness map.

The loan that develops a player and abandons the ownership

One thing needs saying here, because domestic cricket conversation drifts toward big clubs and franchises.

In Bangladeshi domestic cricket we sometimes see a large club take a young player on loan, play him for two or three seasons, then release him without any commitment. If a club borrows a player, plays him for three years, and never once takes ownership of him in those three years, what exactly is it producing? Not an improved player. It is producing the market value of someone who may then not even be given the chance to play on his own account.

The small clubs' arithmetic collapses exactly here. They invest in coaching, physio and match experience, and someone else lifts the reward into a trophy cabinet. This is not a moral complaint; it is an observation about an equation. The transfer market is a rumour engine with a settlement date — and in Bangladeshi domestic cricket that date frequently never arrives. A contract that never settles is not a contract, it is an injunction. And the people living under an injunction are the ones most durably labelled "developing," without ever being given the chance to be built.

What I am watching now

From November to February, what happens on Bangladesh's domestic grounds sets the shape of the national side two years later, while remaining almost invisible in the news cycle. Because nobody watches in Khulna, what happens there is not recorded — and because it is not recorded, it is treated as though it never happened.

Three things I am tracking separately this season.

First, over-load. Who is bowling more than sixteen overs in an innings, and how old are they — that list is next season's injury list, just written earlier. For a seamer like Hasan Mahmud the arithmetic matters twice over: squeezed by the international calendar, and simultaneously carrying his domestic side's load almost alone.

Second, spell-level accounting rather than wicket counts. Everyone will know the tournament's leading wicket-taker's name; the real picture belongs to the bowler who bowled the second session of the innings. A scorecard does not understand sessions, only sums.

Third, the new window. How many players under twenty-three get eight or more matches this season, and how many disappear after four — that ratio will set the shape of the side for the next four years. To my knowledge, nobody is calculating that ratio.

None of these three questions will be answered in a podcast, a stream, or a fantasy platform. They have to be written down by hand in a gallery in Khulna — the dataset has to be built, and then the question has to be rewritten. The numbers will accumulate year over year, and eventually a sentence will stand. By then the reader will learn that the story we told for so many years was never a story about cricket; it was words written in the margin of a blank scorecard. The only question left is this: next season, will anyone read the margin?