The Craft of Burning Reviews: Bangladesh’s DRS Ledger and the Unfinished Win-Probability Accounting Under Tournament Pressure

The last five overs required slightly more than six runs an over. A rain-shortened chase of 114 in 19 overs is not, by the arithmetic of modern T20, an unreasonable ask. Bangladesh batted that innings to an eight-run defeat, with an opener's score as the highest entry on the card. In my match ledger, the most expensive event of that night was not a short ball or a yorker. It was a review — and the accumulation of review decisions taken in the overs before it.

I have watched this game for more than fifty years, and since 2026 I have held to one rule: before I quote a number, I re-watch at least ten matches myself. That rule has made me a slow columnist, but it has given me one thing — when I say a review was burned, I say it from my own log, not from a highlights package. Across Bangladesh's three Super Eight matches of the 2026 T20 World Cup I recorded every review, every soft signal, and every third-umpire decision separately. This piece comes from that ledger.

Context matters more than usual here. The 2026 format produced a strange thing: group matches in New York and Dallas, where the pitch and outfield behaved one way, and the Super Eight in Antigua and St Vincent, where they behaved another. The same squad had to change its batting order and bowling plan three times in three weeks. New York offered a pitch where the ball held; Antigua offered pace onto the bat. Teams that treated travel as a tactical input gained an edge. Teams that assumed T20 is the same everywhere carried their group-stage arithmetic into conditions that had already changed.

Bangladesh beat Sri Lanka, the Netherlands, and Nepal in the group, and lost to South Africa by four runs. In the Super Eight they lost to Australia and India, then to Afghanistan by eight runs. The summary is simple and unflattering: against the bigger sides they competed and did not win; against Afghanistan, where the opportunity was clearest, the match cooled in the last five overs.

None of this is new. Bangladesh finished eighth in the 2026 ODI World Cup with two wins, against Afghanistan and Sri Lanka, and lost the 2026 Nidahas Trophy final off the last ball. A pattern is forming across a decade: the group-stage accounting is sound, and the knockout accounting slips one step behind. This article is an attempt to find the cause — and that search begins with the review ledger.

My ledger has three separate layers. The first is the DRS layer: who reviewed, in which over, under what circumstances, and what the outcome was. The second is the model layer: win probability and expected runs, and what the model was saying at which point. The third is the eye test: what I actually saw — the batter's footwork, the bowler's release point, the keeper's position, the field changes.

I keep the layers apart because each has a claim it can legitimately make and a claim it cannot. The DRS layer can measure consistency of decision-making; it cannot measure correctness, because the final call belongs to a technology that has its own uncertainty. A win-probability model can express likelihood, not causality. The eye test can judge technique, intent, and pressure; it cannot predict. Analysts who blur the three usually use none of them well.

I opened the expected-runs trap and found the eye test still admissible — but only within a narrow jurisdiction.

The common complaint about Bangladesh's review use is that they review too often, emotionally, and burn the review in the first over. My ledger says that is partly true and not the whole picture. The problem is not the number of reviews; it is the order of reviews. The review with the highest probability of success is not taken; the review with the lowest probability is. Under tournament pressure this ordering error grows, because pressure makes people compute feelings rather than probabilities.

On the field it looks like this. The projection knows about swing and seam. The fielder at slip does not. The batter knows whether he missed the ball; he does not know whether it struck pad or the inside edge. Decisions are taken inside that information asymmetry. Teams that decide from the dressing room — keeper and bowler input, one line with the senior batter — do better. Teams that decide from the middle of the pitch on adrenaline do worse.

In Bangladesh's case my ledger shows a specific pattern: a high propensity to review within the first ten or twelve balls of a new batter's innings, especially when the scoreboard is pressing and the side is desperate for a wicket. Psychologically this is coherent — pressure seeks control, and the review is an imagined control switch. Statistically it is the worst possible window, because pad-bat uncertainty is highest against a new batter. The review gets spent when it is least likely to succeed, and is unavailable at the end when each decision weighs most.

A precedent belongs here, along with its limits. When player reviews arrived in 2026, the rules were entirely different — soft signal, a three-umpire panel, and different retention rules. After 2026 the ICC allowed retention on umpire's call in LBW, and the arithmetic changed substantially. I do not write both eras in the same ledger. A wrong decision under the old rule may not be a wrong decision under the new one. Precedent works only when era, format, technology, and conditions are written beside it. Otherwise it is nostalgia, and nostalgia makes decisions easier by making them simpler. If you assume Bangladesh always loses the pressure match, you stop looking for the cause of the Afghanistan defeat and start collecting confirmations.

On win probability: the model had Bangladesh ahead at several points in those last five overs. It jumped on every boundary and fell on every dot. That oscillation is what I trust least, because the model does not know who is batting, who is bowling, who is tired, whose hands are steady. It knows score, balls, wickets. I let it answer two narrow questions: historically, what share of sides win from this position, and what has the run rate been over the previous ten balls. The third question — what this bowler will do to this batter on this pitch — is outside its jurisdiction. That belongs to the eye test.

And the eye test deceives too. Humans love patterns, even where none exist. Three similar dismissals become a technical flaw. That is as wrong as drawing a tournament conclusion from one win-probability curve.

I learned that most clearly in 2026. When football returned to empty stadiums in May, I logged fifty-five matches and found home advantage was not only about crowd noise but about referee decisions. In cricket I ran the same test on a different question: does home advantage fall without a crowd? The West Indies series in England in July 2026, the first international cricket after lockdown, suggested that what disappears is not batting skill but bowling rhythm. Without noise, the signal that a bowler has lost his run-up arrives later. Bowlers were late; batters benefited. In the Caribbean in 2026 the crowd was present but it was not Bangladesh's crowd — a half-version of home advantage, in which bowlers appeal more and fielders press for reviews with more confidence. The noise variable must always be logged as three things: whose noise, how much, at which moment.

Selection is the second thing that breaks under tournament pressure. A general rule in my ledger: a winning XI deserves continuity — but only while conditions are continuous. New York to Antigua changes the pitch, the ball, the light, and the dew. The XI that was right in New York may not be right in Antigua. Bangladesh's squad carries a structural imbalance — few genuine all-rounders — so adding batting subtracts bowling. Under pressure, coaches and captains reach for the safe pick, and the safe pick is the experienced name who may not be in form. That is ledger bias: the ledger remembers the old name and forgets recent form.

Star load management arrives differently for Bangladesh. For wealthy sides, resting a star is a luxury; for Bangladesh it is close to impossible, because squad depth is thin. The consequence is that the star plays every match, and by the back end of a tournament he is tired. That tiredness is invisible on the scoreboard and visible in footwork — and bad footwork produces bad reviews, because a batter who arrived late knows he arrived late and reviews out of guilt. A review error and a footwork error are not two problems; they are two faces of one fatigue.

On pitch maps, I use them as a limited witness. They can say where the ball landed; not what the batter intended. Against Afghanistan, the balls that troubled Bangladesh in the last five overs were not yorkers or slow bouncers. They were back-of-length, outside off, where power is unavailable and rotation carries risk. That length earns no highlight and wins matches, because it forces a batter to confront his own limits. Models sometimes mark such overs as neutral, because runs fell without wickets. I call them pressure overs: invisible on the card, causal for the six that follows.

The same applies to spin matchups. Bangladesh's spinners found help in New York and less in Antigua, and the difference was flight as much as surface. A spinner who releases from above benefits on a holding pitch; a spinner who pushes the ball through is defenceless against the sweep. No matchup graph shows that. Only the release point does.

The dominant post-tournament narrative is about courage: the side fought, the side did not lose its belief, the side stayed with the big teams. That narrative is not false, but it is incomplete, and incomplete narratives are dangerous because they teach the wrong lesson for the next tournament. My ledger is less dramatic. The side fought where the tempo of the match was in its favour and lost where tempo was controlled by the opponent. The difference is not talent; it is control, and control comes from two places — the death-over bowling plan and the management of the middle-over run rate.

A counter-argument deserves its hearing, as the ledger rules require. The opposing case: the real problem is skill, and reviews and footwork are symptoms. That cannot be dismissed. Bangladesh's batting average and strike rate in the 2026 ODI World Cup trailed the leading sides. My answer is that both are true in different proportions, and the proportion is format-dependent. ODIs expose skill gaps because the match is long. T20 exposes management gaps because the match is short and a handful of decisions settle it. Losing a T20 tournament on management is the better news, because management can be taught in months and talent takes a decade.

A second objection is stronger in raw numbers: a review affects at most two or three decisions in a match of 120 balls. True, and irrelevant to weight. A review converts directly into a wicket, and a wicket in T20 is worth ten to fifteen runs. Ten to fifteen runs is usually the match.

A third objection is the most uncomfortable: perhaps Bangladesh's review management is not poor, but the technology's uncertainty is large. Ball-tracking error margins and the umpire's call boundary — less than half the ball projected to hit the stumps — are built so that clear errors are sometimes retained. I accept this and do not accept it entirely. Technological uncertainty is equal for all sides; review skill is not. Unequal outcomes do not permit equal causes.

The errors form at three levels. Information: the projection is not visible to the people on the field. The fix is technical — a short video confirmation before a team review — and the ICC has not adopted it because it slows the game. In T20, two minutes of delay costs less than one wasted review. Process: many sides assign the decision explicitly — captain decides, keeper inputs, bowler may veto. In Bangladesh that process has been unclear; sometimes the keeper pushes, sometimes the bowler, sometimes the captain decides alone. Ambiguity slows the decision, and a slow review is usually a wrong review. Culture: in South Asian cricket the review is often treated as protest — I believe I am not out, therefore I review. That makes the review an act of self-defence rather than a search for truth. Sides that treat it as a strategic asset decide in advance which phase of the match the review is most valuable, usually the last four or five overs. That requires restraint in the first ten, and restraint is the scarcest resource under pressure.

The most important decision for Bangladesh at the next tournament is not an XI. It is a written review protocol: who decides, after which over a review may not be used, and which situations never justify one. It is not glamorous work. It is the kind of work that stays in the ledger. In fifty years I have not seen Bangladesh lose for lack of talent. I have seen them lose for lack of procedure — the kind that protects people from guessing. A side that moves a review decision out of five seconds of emotion and into ten seconds of process saves eight to ten runs a tournament, and eight to ten runs is usually a match. Watch the first ten overs next time. If no review is taken there, the process is working. If one is, what changed was the shirt, not the ledger.