The Home-Win Number Is True; the Explanation Is Not

At the Sheikh Abu Naser Stadium in Khulna, I hand-coded the scorecard of a National Cricket League match from last season. Three hundred and eighty-seven deliveries, two innings, not a single highlight clip, no digital scorecard — only a printed sheet and two lines in a local weekly. Building the table ball by ball, one number stood out. The winning side's three spinners bowled seventy-one per cent of the match's overs; the losing side's two seamers bowled eleven overs in the second innings, six of them in an unbroken opening spell.

Nobody has written about that Khulna pitch. Yet in the safe databases of Dhaka, first cousins of that seventy-one per cent return every home series. In nearly every Test Bangladesh has won at home across these seventeen years, the spinner share of overs sits above sixty-four per cent. Still a question stands in place: is seventy-one per cent a cricket fact, or the result of sampling?

The numbers were not lying. They were waiting for a different question.

From 2026 to 2026, I coded every home Test session by session. Not ball by ball — footage does not exist for every match; scorecards, weekly sports pages, and whatever I watched myself. Let me be explicit at the start: every figure in this piece comes from a table I built by hand, not from an official database. If there are errors, they are mine. The sample looks bigger than it is.

Across those seventeen years, Bangladesh played fifty-six home Tests. Sixteen won, twenty-six lost, the rest drawn. Roughly a twenty-nine per cent home win rate. Away from home in the same period, they played more than sixty Tests and won a handful.

Put the two numbers side by side and a story forms: Bangladesh are strong at home, weak abroad. The story is true. The story is also a beginning, not an analysis. Because inside those fifty-six matches two separate populations are hiding, and we usually merge them into one average.

Split the fifty-six and the arithmetic changes. In my table I divided the home Tests into two groups. Group one — Australia, England, India, South Africa, New Zealand: the top five. Group two — Zimbabwe, Ireland, Afghanistan, West Indies, Sri Lanka, Pakistan.

In group one there were nineteen home Tests and two wins. In group two, thirty-seven and fourteen wins. Roughly ninety per cent of what we call ‘home strength’ comes against teams outside the top five. That is not a claim about cricket; it is the structure of the table.

Now the spin accounting. Against the top five at home, our spinners delivered an average of fifty-eight per cent of total overs. In group two, that rises to seventy-two per cent. The character of the pitch does not shift that much; what shifts is the baseline of the opposition batting and the way our spin stock is used. Only nineteen home Tests came against the top five — in a sample that small, one series of rain, one toss, one spinner's one innings can move the whole percentage.

The most valuable number for me arrived where the counting does not: in results that never happened. Of the fifty-six home Tests, twenty-three lost significant time to rain. In nine of those, Bangladesh led on first innings and still got no result. Six of those nine came against group-two sides. Where the probability of a win was highest, the sky cut the account.

In Khulna I learned that silence is also a dataset. A match that never became one is still part of our table — we just do not write it in the column.

Then the toss. In home Tests where Bangladesh won the toss and batted, the average first-innings score was 264. Where they lost the toss and fielded, the opposition's average first innings was 387. The gap is enormous, but the explanation is not luck — it says that building a Test for the fourth innings is our plan; and when the opposition bats first and posts a big score, that plan has no answer in our hands.

There is another layer the ordinary home-series table almost never shows. Look at the geography of recording across those fifty-six matches. Matches in Dhaka or Chattogram had at least one camera; session-level data could be built. Matches treated as back-up fixtures at other venues have no digital archive. Which means our dataset called ‘home performance’ is really a sample of camera placement. Where there is a camera, there is data; where there is data, there is a selector's eye. This is not a conspiracy, it is sampling.

Apply the same logic to home spinner workload and a different picture emerges. In home Tests our frontline spinners average forty-four to forty-eight overs per innings; for a bowler of Taijul Islam's type this is not unusual, but the model of that load is imported from European and Australian conditions. There, spinners rarely bowl more than thirty overs in an innings. So our spin-first plan works at home, yet haul the same bowler beyond five or six Tests a year and the curve steepens sharply.

Reading home-series numbers requires one more thing in mind: opposition selection is not a neutral process. Future Tours Programme windows are set long in advance, under marketing, broadcast and travel-cost arithmetic. Bangladesh's home window contains fewer top-five Tests because the big boards' home calendars are full. That scheduling reality manufactures our win rate — not the pitch, not the rankings, not preparation.

The obvious explanation is that we prepare spinning tracks at home, therefore we win. Preparing a track and winning are two separate events, and the relationship between them is less simple than it looks. Evidence: in home Tests where the opposition brought at least two experienced spinners, our spin advantage almost vanishes. In twenty-eight such matches in my table, the opposition spinners' share of wickets equalled or exceeded ours.

The pitch cuts both ways, but the benefit of the cut belongs to whoever's batting technique can take it. Our spinners bowl more overs; part of that is simply that home series often field two seamers instead of three. And yet the rate of batting collapse is nearly identical home and away — in both places one innings in four ends under two hundred. That number says the home wins have come more from exploiting the opposition's spin weakness than from a structural improvement in our batting.

My hypothesis was that pace-avoidance caused the home success. The data did not reject it outright, but it changed the explanation: pace-avoidance is not the cause of success, it is the condition of it. The actual cause is the joint product of opposition selection and toss planning. This is where the boundary between correlation and causation becomes visible.

Every model is a prayer until the data says otherwise. What it said here is not comfortable.

The phrase ‘golden generation’ needs the same treatment. Whether a generation is golden is not measured in trophies; it is measured on the peak curve. An all-rounder of Shakib Al Hasan's type typically peaks between twenty-eight and thirty-two; a batsman of Mushfiqur Rahim's type between twenty-seven and thirty-one. When those innings coincide, a window opens — and if that window gets few series against the top five, the valuation of that generation rests on incomplete data. The unwatched scorecard in Khulna and the Mirpur highlight reel will not reconcile unless both are laid on the same table.

The recurring argument over workload management for Taskin Ahmed and Nahid Rana rests on the same foundation. At home, spinners absorb overs and seamers absorb spells; but what follows a home series is usually an away tour. Pace load is therefore not a per-Test accounting but a per-calendar-year one. Judging from a single series is not reading the map, it is recognising a road by walking one step of it.

For the next home series I will watch three things, and I have already built the tables for all three. One, the opposition list — if the top five arrive, win rate cannot measure the team's improvement. Two, the accounting of rain-lost sessions — the more matches that return unfinished, the more incomplete our table, and those blanks are the next question. Three, toss planning — building a fourth-day pitch and fielding on day one cannot both survive as decisions.

Perhaps that is the real return. The number never lied; we were simply asking it a question it could never answer.