The Fourth-Innings Collapse: The NCL Number Nobody Counted

Last February, at Bogra's Shaheed Chandu Stadium. First session of the third day. A left-arm spinner from Khulna Division was bowling a long spell; beside his name the scorecard read 24-3-51-4. My notebook, beside the same spell, says something different — three of those four wickets came from a batsman's front pad locking up and a simple catch at square leg, not from turn off the pitch. Sitting at grounds in Khulna taught me this: the truth of domestic cricket never fully makes it into the scorecard's columns, and whatever fails to make it there tends to become the strongest evidence later. But the real number from that morning was not the spell. It surfaced after I hand-coded 68 first-class matches across three seasons: of the games in which a side took a first-innings lead, only 41 percent were won. In first-class cricket that rate sits well below the normal average. In Bangladesh's domestic four-day game, a first-innings lead is behaving almost like a coin toss — even though the whole architecture of the format is built to reward it. Method first, because however striking the number looks, this is a scorecard-level dataset, not ball-by-ball. The National Cricket League is Bangladesh's premier first-class competition. Eight sides — Dhaka, Chattogram, Khulna, Rajshahi, Barishal, Sylhet, Rangpur and Dhaka Metro — play four-day matches. Most games fall in the winter window, October to December, sometimes spilling into January and February. The venue list includes Khulna, Rajshahi, Bogra, Sylhet, Cox's Bazar and Chattogram. A large part of this competition never reaches television. There is no ball-by-ball log, scorecards are entered late, and some innings remain incomplete in the record. That is where much of my work sits — hand-entering scorecards, taking notes off radio commentary, and sitting at grounds in Khulna to watch the cricket myself. This log of 68 matches from 2026-22 to 2026-24 is the basis of everything below. One more thing has to be understood before reading this dataset: when the NCL is played. The competition often overlaps with the BPL window or an international series. In the very weeks the NCL runs, the country's best pacers, best spinners and top-order batsmen are frequently away on national duty or in the BPL. What we call the NCL is, in effect, a filtered NCL. I wrote the hypothesis down first, because inventing a story after seeing the numbers is easy. My expectation was that a first-innings lead in the NCL would be close to decisive, that the leading side would win more than 60 percent of the time. The logic was simple — winter pitches dry out, spin grows on days three and four, and the side with a lead gets to set traps with the turning ball. The result contradicted the expectation directly. It is worth saying that the NCL is a strange archive of Bangladeshi cricket. Taijul Islam's left-arm spin, Mominul Haque's patient batting, the early days of Shakib Al Hasan turning out for Khulna Division — the roots of all of it are in this competition. Yet this is precisely where the least data gets recorded. The trophy is common here; the archive is rare. A first-innings lead converts to a win 41 percent of the time — and breaking the number down reveals three separate layers. The first layer is venue. At the north-western grounds (Bogra, Rajshahi, Khulna) the leading side wins 48 percent of the time; at the Dhaka, Sylhet and Chattogram venues that drops to 36 percent. Where the pitch is supposed to turn more, the lead is worth a little more; where batting is easier, the lead is nearly worthless. The second layer is the fourth innings. Of the matches that reached a fourth innings, roughly 62 percent ended under 200; and by median rather than mean, the figure is sharper still — 148. Something odd shows up here. Most matches whose fourth innings passed 200 ended in draws, and in almost all of them one pattern repeated: a set batsman survived a long time at one end while wickets fell at the other. The third layer is the distribution of wickets by innings. In my log, spinners took 54 percent of wickets in the second innings, rising to 68 percent in the fourth. The pace share falls the other way. On the surface this supports the story of home spin dominance. But the same dataset shows something that weakens that story. Spinners are taking wickets, yes — but the number that matters is how many overs they are bowling. In the fourth innings, a leading spinner bowled an average of more than 31 overs. That is not the product of an attacking plan; it is the product of necessity. Because the pacers had so little rest earlier in the match, the fourth-innings workload fell almost entirely on the spinners' shoulders. This is where scheduling enters. NCL matches are packed into a very short window. A side sometimes plays two matches four or five days apart. At that turnover, a young pacer who bowls 35 to 40 overs across four days loses both his pace and his edge on day four. The captain is left without a trustworthy pace option in the fourth innings, and the spin share of wickets inflates artificially. In my log, among pacers aged 19 or 20 who bowled more than 35 overs in a single match, the rate of a major injury in the following season was clearly higher. There is a gap between a body's peak curve and the demands of the competition, and that gap is being filled by loading it onto young shoulders. A young quick's greatest asset is his youth, and the domestic structure is using that asset with the least care of all. One more thing kept returning in my log: the toss. The side that won the toss chose to bat first 71 percent of the time. Yet the toss-winning side won only 53 percent of matches — meaning the toss advantage barely translated into results. The toss is a decision, not an outcome, and the two are routinely confused in the discussion. This is where the dataset's biggest gap becomes obvious — the thing I cannot count. My 68-match log has no ball-by-ball. So I do not know what the fourth-innings collapse was built on — the pitch, the breeze, or the batsman's own head. There is no pitch report, no injury log, no written reason for selection. Three scorecards were wholly incomplete; I dropped them, but dropping them is itself a decision, and that decision carries a cost in my results. I tried to build a control group — only the NCL matches that were streamed or for which a detailed log exists. The sample shrank so far that no conclusion could survive it. That is also a result, and it is my most honest one: the questions that matter most about Bangladeshi domestic cricket cannot currently be measured with the data we hold. Now the question that, if skipped, turns this whole piece into a trap: between a first-innings lead and the match result, is there a relationship, a cause — or are both the shadow of a third thing? My suspicion is that the sentence 'spin rules domestic cricket' is partly a measurement artifact. For a few reasons. First, the scheduling window. Almost every match is played in the driest weeks of the year. The behaviour of the pitch is a product of the calendar, not of competition design. That is a truth about the sample, not about the cricket. Second, venue selection. Where a ground has a good curator, the pitch holds up for three or four days; where it does not, the surface breaks on day two. In my log, match lengths at several north-western venues were clearly shorter. That is not a story about spinners' skill; it is a story about groundstaff and budgets. Third, survivorship bias. Matches whose scorecards were never fully entered are absent from this dataset. Perhaps those were exactly the games in which the pacers produced a different outcome — we do not know, because we did not see them. I do not chase edges; I build a monastery around them. So, honestly: this 41 percent is not final proof — it is an indicator, a direction. The numbers were not lying; they were waiting for a better question. And in Khulna I learned that silence is also a dataset. Next season I want to count three things. One, the average fourth-innings score — if it starts passing 200, then scheduling or pitches have changed. Two, the overs bowled per match by pacers under 21, and their availability the following season. Three, the matches whose scorecards stay incomplete — because the data that never gets recorded says the most of all. The spike got spiked, but the pattern stayed in the data. The question now is not whether spin rules the NCL; the question is whether we are measuring that rule at all, or merely counting its shadow and writing the story from that.