Empty Stadiums, Full Screens — Why the Field of Data Stayed Empty

I still remember that afternoon at the Sher-e-Bangla Stadium in Mirpur. During the pandemic years the galleries were empty; nobody sat in the green chairs, and instead of the roar of a crowd the loudspeakers carried only the drone of a compressor. Two cameramen stood beside the pitch, and a thin line of dust drifted under the floodlights. In that silence a question lodged itself in my head — if we don't watch the game, who makes the decisions about it?

Because at that same moment something else was empty, and nobody could see it: the analytical table. A scorecard, a highlights reel, a press conference — those three were, for a long while, almost the entire raw material of cricket discussion in Bangladesh. When the input is empty the output is empty too; that is not philosophy, it is the plain rule of any pipeline. And that gap is the least-discussed structural problem in Bangladeshi cricket.

The Bangladesh Cricket Board is an institution with three decision centres — board-level policy, the selection panel, and franchise ownership. Each is an incentive map. Who stays in the squad, who rests in which format, who sets the television schedule — the answers are never written only in the performance ledger. Reading that ledger for more than two decades, one thing has returned again and again: a decision taken inside the board surfaces on the field two to five years later, and the delay is the story — not the controversy.

The pandemic inverted this timeline. In mid-2026, with the Sher-e-Bangla empty, the centre of decision-making drifted from the board to the screen. I watched the pandemic empty the stadiums, then fill the screens. Streaming, YouTube, Facebook Live — cricket suddenly sat down in a different room. Revenue models changed, schedules changed, and most importantly, who gets to watch changed.

Leave out broadcast economics and the picture is incomplete. In the streaming era a match's start time is set not for the audience's convenience but for the platform's algorithm and its advertising slots. Domestic cricket in Bangladesh lands in the weakest part of that arithmetic — where attendance is low, investment is low, and therefore the staffing for data collection is low. The circle is hard to break, because every party waits for another to invest first.

Here lies a curious paradox. The screens filled up, the audience grew, and yet the field of data stayed empty. An institution that never built a systematic data pipeline is suddenly required to supply far more content — every match, every series, every innings. The result? Faster decisions resting on slower evidence.

Take selection. A batter is scoring heavily in domestic cricket. The question sounds simple: should he be picked? The better question is harder — in which situation, against which ball, at which over? His strike rate in the first ten overs is not his strike rate in the last five; his average against left-arm spin is not his average against right-arm pace. If those splits are not on the table, selection becomes a vote of memory and impression. The decision is made, but its foundation stays empty.

I first began to understand this gap through football, not cricket. In 2026, after launching a tactical newsletter called "Half-Space Notes," I analysed RB Leipzig's 4-2-2-2 against Dortmund and found that counting Naby Keïta's twelve ball recoveries in a single match made the team's pressing map suddenly legible. Numbers alone do not change decisions, but without numbers decisions go blind. Returning to cricket, I understood the same rule applies.

So when someone says Bangladesh's bowling attack collapses "for lack of experience," I pause. Because that sounds more like habit than analysis. The real question is whether anyone holds the map of which batter scores against which bowler, in which over, under which field setting. At the 2026 Qatar World Cup, Morocco played a 5-4-1 low block against Spain; Sofyan Amrabat made ten ball recoveries and four tackles in one match, and across seven matches Morocco conceded only five goals. That block was not a wall; it was a timed counterattack — every gap counted in advance, every passing lane closed in advance.

This is defensive block geometry — and to my mind it is the most underused tool in Bangladeshi cricket. We think of bowling in terms of wickets and of the field with our eyes. In modern cricket, though, a bowling plan is geometry: which zone narrows a batter's options, where a fielder cuts off the scoring lane. Drawing that map needs only two things — over-by-over data, and the patience to read it. With that patience, even a weak bowling unit can stand inside a structure.

Patience is the real shortage. Bangladeshi cricket loves change — coach, captain, format, opening pair, nothing stays settled for long. We treat each change as a turning point. But I have learned to ask first whether the incentive structure has actually moved. If it hasn't, that is not change, it is variance. And mistaking variance for signal is the most expensive error in analysis.

In 2026, I was on radio commentary for the Bangladesh–Kenya match at the ICC Trophy. Decisions then were made the way they often still are — on thin information, under strong emotion. Thirty years on, the technology has changed and the platforms have changed, but the volume of information in the decision room has not grown much. The pattern was already there before the whistle blew; we merely looked for its explanation afterwards. This is why compounding and coincidence must be separated. At the 2026 World Cup in Russia, Mbappé ran at 37.1 km/h against Argentina; that was no coincidence, it was the compound product of speed built over a year. In Bangladesh we often read one moment of outcome and infer the cycle — backwards.

I began to see transfers as bets on human adaptation; franchise auctions in cricket are no exception. Signing a player means buying not just his average but his body, his age curve, his capacity to adapt to a new environment. Placing that bet without data is rolling dice with your eyes shut. Yet in our auction talk, averages and highlights are the chief witnesses, while splits and workload are almost absent.

On workload, one point matters. At the Tokyo Olympics in 2026, Pedri played six matches for Spain, and I warned then about where that load would land a teenager later. The same question applies to Bangladeshi fast bowlers. I learned to read injuries as data points and recoveries as tactical choices — who bowled how many overs, returned in how many days, returned in which format. Granting a place without reconciling those three numbers means borrowing an injury from the future.

Now the counter-question. Suppose that from tomorrow a full data pipeline runs in every series. Would the problem be solved? My doubt says no — because a governance choice sits behind the empty table. In a structure that prizes loyalty above evidence, more data only produces more argument; it does not change the decision. I also accept that data itself is a kind of intoxication — the algorithm became the scout before the scouts noticed, but what the algorithm says is often not the answer to a question, only a repetition of it. When numbers are in hand, there is a temptation to cite them; for every dataset one must ask which decision it changes. If the answer is "none," cut it.

There is another trap, and it lands hardest on analysts like me — institutions crowd out the human game. Contracts, governance, broadcast deals are easy to write and easy to source. But a sound from the stands, a person standing near the boundary, the light at five in the afternoon have no source, and yet analysis goes bloodless without them. So I keep at least one thing in every piece that cannot be placed on a table.

So what should you watch in the next match? Look at Bangladesh's bowling changes in a specific series and ask whether a genuinely new over-by-over map lies behind them, or whether this is only variance. My forecast is this: of the changes you see over the next two years, most will be unrelated to results. And I leave an easy path to prove my read wrong — if any selection panel begins publicly deciding on split data, then I will say the field is no longer empty.