From the Khulna Scorebook: Spin Share, Pacer Workload and the Misreading of Day Four in the National Cricket League

From the Khulna Scorebook: Spin Share, Pacer Workload and the Misreading of Day Four in the National Cricket League

A December afternoon at the Sheikh Abu Naser Stadium in Khulna has lodged itself in my notebook. 2:40 pm, the 22nd over of the third innings. A left-arm spinner bowling his fourteenth consecutive over from the pavilion end, the board reading 61 for 4. Nothing in that over turned sharply; nothing bounced unexpectedly low. What happened was crueller than that — the batsman already knew roughly where the next ball was coming, and still could not deal with it.

Back at the hotel that evening I opened my ball-by-ball sheet. Across the 47 NCL innings I have logged over the last three seasons, that afternoon was not an exception. It was a sample of the rule. After the 20th over the spinner's share of overs climbs, the fast bowler's spells shorten, and the match slows to a walk. The real problem in our domestic first-class cricket is not the pitch; it is the decision structure that determines who bowls when.

The National Cricket League is Bangladesh's premier first-class competition. Eight teams take part — Dhaka, Dhaka Metropolis, Chattogram, Khulna, Rajshahi, Rangpur, Barishal and Sylhet. Four-day matches, a season that usually runs from October into December. The Sheikh Abu Naser Stadium is Khulna's home ground; it hosted an international Test in February 2026.

Here is the difficulty: ball-by-ball data for this league is not published anywhere. Scorecards exist — who scored how many, who took how many wickets. But which bowler bowled which over, in which spell, behind which field, is recorded nowhere. So almost every judgement we make about domestic cricket rests on three numbers: average, strike rate, economy.

Since 2026 I have logged it myself. Nine data points per delivery — bowler type, spell number, line, length category, shot, shot direction, field position, runs, and whether it was a wicket. Forty-seven innings works out to roughly twenty-five thousand balls. It is not a large sample, and I never claim that the Khulna press box can read the whole country's cricket. But this much is true: the questions this log raises are questions a scorecard cannot even formulate. The spreadsheet was my prayer mat; the data, my daily office.

Not the pitch — the budget

By "spin over share" I mean the percentage of total overs bowled by spinners. Across 47 innings it is 63.4%. That number alone says nothing, because it is a whole-match average. Broken down by hour, the picture inverts.

In the first hour of play, pace takes 58% of the overs and spin 42%. Before the morning session is over it has flipped — pace 39%, spin 61%. On the second morning some balance returns, with pace at 44%. But across the third and fourth innings combined, spin's share is 74%.

So teams open with pace and switch to spin the moment the first hour is gone. Why?

The familiar answer is that the pitch is spin-friendly. My log does not support it. Deliveries that skidded low or turned more than normal account for 21% of the innings; the other 79% arrived at ordinary bounce. The pitch is not breaking up, and yet teams lean further and further into spin. Something else is doing the damage.

It is the arithmetic of run cost. In the first hour, pace goes at 3.4 an over and spin at 2.6. In the first hour, a seamer takes a wicket roughly every 11 overs; a spinner every 14. Pace buys wickets quickly but expensively; spin buys them slowly but cheaply. A captain trying to hold a first innings under 350 will naturally trade expensive overs for cheap ones. That is not tactics. That is budgeting.

And the budget cuts the wrong way. A spin share above 60% means the opposing batsman faces sixty or seventy overs of the same type of delivery behind the same angles of field. Past the thirtieth over the spinner is no longer trying to take a wicket — he is waiting for the batsman to make a mistake. In domestic cricket batsmen will make mistakes, because nothing else is being offered to them.

Field geometry is part of this arithmetic too. In the first hour, with pace on, point often sits inside the circle and square leg is up. After the 20th over the picture changes — deep point, deep cover, long-on. The side is no longer hunting wickets; it is plugging runs. The real contest therefore happens not between bat and ball but in the empty pockets of the thirty-yard circle, where nobody is standing. Who fills those pockets, and who deliberately leaves them empty, is the least-written story in our domestic game.

Four-over spells, fourteen-over sentences

Across 47 innings, the average pace spell is 4.2 overs. For spinners it is 13.7. In a day a seamer bowls about 15.8 overs; a spinner bowls more than 34.

That sounds ordinary, but it hides something. A seamer bowls four overs, stands near the boundary for ten minutes to cool down, then comes back. In my log, the rate of wides and no-balls in second spells is 38% higher than in first spells. The reason is not complicated — the line shortens in the second spell and the rhythm is gone.

Temperature has to enter the calculation. A December midday in Khulna sits between 28 and 30 degrees with humidity near 70%. Warming up and cooling down sixteen or seventeen times in a day in that environment is not easier than a spinner's long unbroken spell. It is probably harder. I trust the model, but I audit the story it tells. The evidence for spell length is not in any single match; it is scattered across twenty-five thousand balls. This is inference, not conclusion.

The economics of the first-innings lead and the draw

The NCL's points system rewards a first-innings lead. That is the league's most powerful behavioural driver and its least discussed.

The result shows up like this: a side that reaches 300 in the first innings stops trying to reach 450. At 310 or 330 they declare. The opposition then bats in avoid-the-follow-on mode — not scoring, just consuming time. In the matches in my log where third-innings follow-on pressure existed, batsmen struck at 38.2, fourteen points below the league average. Fifty-eight per cent of those matches were drawn. In matches where the follow-on was never a question, a result arrived 71% of the time.

The behaviour is not irrational. A draw beats a defeat. The question is whether the side is refusing to chase a win or simply afraid of losing. From a press box the two are hard to separate, and here I will not push past the model, because I have no instrument for measuring fear.

How do you convert a domestic average into a Test average?

Selectors read NCL averages. But what is the average actually saying?

Last season, six of the top ten run-scorers in the NCL had a strike rate under 50. To survive Test cricket you need 50 to 60, but the two environments differ completely in ball speed and pitch character. Domestic cricket delivers the ball at 75 to 78 miles per hour on relatively even surfaces. Test cricket delivers it at 82 to 88 on skeletal or damp ones.

In other words, we have no formula that converts a domestic average into an international one. What is missing is an adjustment model — something football's transfer analysis worked out fifteen years ago. In cricket we still operate on the assumption that whoever does well at home should simply be promoted. Every selection is a prior; the league is Bayesian theatre, and we keep watching the stage while skipping the accounting backstage.

The story everyone tells is that Bangladeshi pitches are spin-friendly, that our batsmen therefore cannot play pace, and that they collapse abroad. That story is worth testing, and my log breaks it in two.

Pitches change; the over share does not. The share of deliveries that turned significantly fluctuated between 18% and 24% across seasons. The spin over share held between 61% and 65% across three seasons. Something that does not move with the pitch cannot be explained by the pitch. It is a decision — the first-hour fear of runs, the short pace spell, the first-innings lead point. Whatever the surface, spin dominance is manufactured from inside the structure.

The second fracture concerns fourth-innings numbers. We say the fourth-innings average is lowest, therefore the pitch has deteriorated. But that sample is contaminated at birth. Pitches that broke in two or three days never produced a fourth innings — and never produced a number. The "fourth-innings average" represents only those surfaces that stayed good for four days. That is selection bias, not a story about the pitch.

One more thing nobody writes down: the side batting fourth has already fielded for two days. A fielder has walked sixty or seventy overs in midday heat. The model wants to explain his decisions as technical skill. But what is deciding in that moment is fatigue, the dawning certainty of defeat, and the hope of one spectacular moment. None of it is measurable, and leaving it out makes the model lie.

In the next round I will count three things. First, whether the spin over share drops below 60%. Second, the wide and no-ball rate of seamers in the first hour, and whether it is higher in second spells. Third, how many matches last into a fourth innings — because that number tells us how much cricket we are actually playing, and how much of it is points accounting.

The press box taught me humility: noise is data too. The question, though, is not a humble one. It is whether our selection system will ever build a model that converts a domestic average into international conditions — or whether we will spend another ten years packing the squad because someone did well at home.