Testimony of Empty Cells: Reading the T20 World Cup Powerplay from the BPL Data Lab

Rangpur, Bangladesh

One night last January I had a spreadsheet open at my desk in Rangpur. The ball-by-ball data for the current BPL season was almost fully entered, but one column simply refused to fill — ball-tracking for the three venues outside Mirpur and Chattogram. Those matches had no bat-swing data, no pitch mapping, no release points. I blinked at the calculator and asked what percentage of the deliveries I actually held were tracked. The answer was seventy-one per cent. The rest was darkness.

Three weeks later, when the T20 World Cup squads and fixtures dropped, I understood that the empty cells were the real story. The question everyone asks before a World Cup — why is Bangladesh's powerplay so slow — had its answer hiding inside incomplete information, and that incompleteness was itself a piece of evidence.

Context: Why the BPL is my laboratory

I have made this claim for years and I still make it: the Bangladesh Premier League is the most neglected T20 data laboratory in the world. The reason is simple. Six teams, two months, a contained tournament, and pitches where 140 and 190 are both possible on the same day. In international cricket you cannot afford a wrong assumption, because every series is a trial. In the BPL you can be wrong, and the wrongness gets logged.

When I moved from cricket writing into the BCB media set-up in 2026, The Daily Star described me as 'the fine cricket writer turned media manager'. Back then I thought writing match reports was my job. Now I understand the job was something else — observing how numbers behave, and asking why they go silent when they do.

In 2026, at forty, I audited rice-mill accounts in Rangpur by day and hand-coded an expected-goals model by night. That was for football. But the method transfers to cricket exactly — put runs where goals go, and wagon wheels where shot maps go. That year, after I published a 132-match breakdown using my own weights, three betting syndicates emailed me within a week. Since then every piece I write follows one rule: every claim carries its sample size, its weighting logic, and a stated error margin. My sentences got shorter; my footnotes got longer.

Core: Three numbers that must be read together

This season in the BPL I logged 10,847 legal deliveries across 46 matches. Of those, 1,612 balls fell in the powerplay, the first six overs. In my model — and this is modelled, not measured — the powerplay run rate came to 8.14, the dot-ball rate 46.3 per cent, and boundaries per ball 17.1 per cent.

Now look at the international figure. Over the past two years, Bangladesh's T20 powerplay run rate in my log is 7.09, with dots at 51.8 per cent. The gap between domestic and international cricket in the powerplay is roughly one run per over, and more than five points in dot balls.

That gap is not about talent. It is about conditions. In the BPL you face domestic seamers whose lengths are often untracked, whose slower-ball frequency lives in no database. Internationally you stand against the same ball when the opposition's analyst unit already knows your powerplay dot-ball pattern. My log shows a clean pattern: Bangladeshi batters are relatively aggressive in the first 18 balls of the powerplay (19 per cent boundary rate), but from ball 19 to ball 36 that rate drops to 14 per cent. That seventeen-ball window is my real subject.

Here I have to admit something. The model was crude. I did not adjust pace separately for each venue, because I had no tracking data for three venues at all. I applied a single generic pitch factor, which is hopelessly simplistic. But the interesting thing is that this simplification is what showed me the empty cells are not an accident — they are a systematic blindness.

Forensics of the empty cells

I opened a blank spreadsheet and let the Bangladesh Premier League teach me, and its first lesson was not diplomatic: the matches without ball-tracking are not randomly distributed. The three venues with no tracking system installed were also the three lowest-scoring venues of the season — my rough estimates are 142.6, 138.9 and 145.1 runs per innings, against an average of 168.3 at the tracked venues.

My dataset therefore structurally under-represents low-scoring conditions. So my computed powerplay run rate reads slightly higher than the true BPL environment. This is not a massive error — probably a discrepancy of zero point two to zero point three. But the direction matters, because the number Bangladesh's planning is built on forgets the country's hardest pitches. When the T20 World Cup hands Bangladesh a slow, two-paced surface, the plan collapses and people say our batters cannot handle pressure. In truth the batters never rehearsed that condition, because by coincidence nobody ever collected the data on it.

When the stadiums emptied — that 2026 stretch — I started measuring what the crowd used to hide. Fielding placements, catch distances, time under throw pressure. That is no longer curiosity for me; it is method. Silence is not zero; it is a new baseline with its own residuals.

The anchor tax: blaming the wrong place

The most popular refrain in Bangladeshi cricket talk is that we do not attack in the powerplay, that our openers play anchors. I tested the claim against this season's data, and the result annoyed me, because it refused to give me a clean villain.

I split the innings into two sets. Set A: innings where one of the opening pair faced at least 30 balls at a strike rate under 120. Set B: everything else. Set A innings averaged 161.4; Set B averaged 169.8. An eight-run gap, which I admit is small with noise around it. But on win rate the picture inverts — Set A innings won 41 per cent of the time, Set B 54 per cent.

So anchors are bad? Slow down. When I looked at the relationship between strike rate and winning, the correlation came out at 0.31 — weak. But when I looked at the relationship between run rate in overs 7 to 15 and winning, it came out at 0.58. And with the last four overs? 0.64.

The powerplay is not Bangladesh's problem. The powerplay is Bangladesh's symptom. The real fracture runs from the eighteenth over to the twenty-fifth ball, that middle window where the anchor is set but the partner keeps changing, and every new batter needs nine to twelve balls to settle. In my log, Bangladeshi sides run at 6.87 in that window, with 44 per cent dot balls, and the most telling figure of all — one single every 2.9 balls, where the BPL's best finisher-rich sides push that down to 1.8.

One thing needs saying plainly. We sell distance covered and high-intensity sprints as effort metrics, but pointless running also produces pretty numbers. If a batter faces 27 balls, eats 12 dots and takes singles off the other 15, his running graph looks superb while the innings has gone backwards. I call this metric inert momentum — end-to-end density where runs do not grow, only balls run out.

Two eyes, two spreadsheets

At Russia 2026 I built a habit that still pulls me along. I watched every match twice — once with only my eyes, and once with a PPDA and set-piece xG overlay in hand. From that habit came a line that later became the foundation of my writing: Germany's press had already decayed before Russia 2026, their PPDA drifting from 8.9 in qualifying to 12.6. Forty thousand people read it. But my model still ranked Germany third-favourite, so I hedged in the text and lost the argument on the result.

That lesson applies directly to Bangladesh's T20 World Cup preparation. If my model blames the powerplay while my eyes see the middle-overs dot-to-single routine, then I have to hold two different theses side by side — and I write that in the footnote of every piece. A model is a monastery: you enter to escape noise, then hear it clearer.

Counter-intuitive angle: there is no anchor, only an empty anchor's chair

Now the place where I make my worst mistakes, and know that I do.

The conventional wisdom is that the anchor batter is Bangladesh's problem. A fairer description: Bangladesh's problem is that nobody exists to keep tempo with the anchor. In my BPL log I found that innings where one batter reached a forty-ball quota at a 110-120 strike rate while the partner held 145-plus averaged 178.2 — meaning the anchor is not the culprit there, the anchor is working, because someone at the other end is holding the pace.

But this is where the danger lives, the danger I know about myself: contrarianism hardens into a brand and pushes me to the opposite of the mainstream. I do not want that. So base rates must be checked. In this dataset 41 per cent of innings had an opener facing at least 30 balls, which is not above normal T20 behaviour — long innings happen in every league. So the problem is not the anchor. The problem is role definition.

In Bangladeshi T20 cricket, no one has clearly defined what a set batter does in the middle overs. So whoever makes 22 off 25 considers himself an anchor, and the team uses him that way. In the BPL, sides that pushed a finisher up early in the middle overs — often at over eleven — ran at 9.84 in the last five; sides that kept a set batter until over fifteen ran at 8.11. Two philosophies, two outcomes, both extant in this league, and no doubt which works better.

The fashion for three spinners or a back-three philosophy now entering our conditions tells the same story. In my log at Mirpur, a batter coming in at number seven faces an average of 11.4 balls — effectively treated as a number four replacement — while striking at 104. We changed the formation but never decided who owns which responsibility.

The appendix nobody reads

I always write on two levels. Outwardly a loud thesis, inwardly a quiet appendix listing everything my model got wrong. This piece's appendix has four names.

One — my pitch factor. I assumed zero point two to point four, probably on the low side. Two — I credited every dot ball entirely to the bowler and gave no weight to the batter's decision. Three — pre-tournament data for the nine matches without tracking, which may pull my variable-based estimate the wrong way. Four — and this is the most damaging — I risk conflating cricket and football metrics here. xG logic and wagon-wheel geometry do not belong to the same family. Before any international comparison I must map domestic match event structures first.

In the last round one Bangladesh XI pushed its strike rate above 125. Next round, my model has sixteen balls left.

Last night I caught a match on a Facebook live stream that has no data in my spreadsheet at all — no coverage, no tracking, nobody recorded it. And yet a number existed there too, written in pencil in the corner of a notebook beside my eye. That number I believe firmly, and that number I can never print in a press box. A spreadsheet, to me, is not the final answer — a spreadsheet is seeing the wrong question clearly.

Takeaway

If Bangladesh bat against an empty cell in the next round of the T20 World Cup, I will not bet on powerplay run rate. I will bet on the dot-ball rate between overs eleven and fifteen — the window no commentator shows as a number. And if dots stay under 38 per cent in that window, then whatever the conditions, Bangladesh will get past 170. Thanks to the empty cells, I now know which good question to ask — and which one I keep forgetting.