The Geometry of Dot Balls: Bangladesh's Chase Dies at Over 10, Not Over 16

June 22, 2026, Sir Vivian Richards Stadium, Antigua. India posted 196/5 — Hardik Pandya 50*, Rishabh Pant 36. Bangladesh's thirteenth over is underway. Open on my laptop is a spreadsheet that logs three numbers after every ball: balls remaining, runs required, and the conditional probability of losing a wicket at that instant. Bangladesh finished 146/8, losing by 50 runs. The scorecard will tell you the bowling failed and the batting collapsed. My sheet had written a different sentence, one I never deleted even after the game ended: this match was not lost in that over. When does a chase actually die? Show anyone Bangladesh's ball-by-ball log from their last five innings and they will say overs 16 to 20. I won't. The over in which a chase truly dies is the over where the oscillation between "score fast" and "don't lose wickets" tilts permanently in one direction. For Bangladesh over the past two years, that has repeatedly happened between overs 10 and 12. Curiously, no wicket usually falls in that window. No big shot is attempted. Only a few dot balls land. And those dot balls are the real date of execution. I built my first xG model in a bedroom in Rangpur, and that taught me to distrust the eye. During the 2026 World Cup I hand-logged every shot in France vs Argentina and built a crude model. The lesson was blunt: what the eye sees and what the event produces are separated by an accounting gap. In cricket that gap is wider, because cricket is not a continuous flow like football. Cricket is a game of discrete events — each ball a separate decision, a separate state. So let me declare the mapping first. In football, xG means the probability a given shot becomes a goal. In cricket that transposition does not hold one-to-one. A single delivery is shaped simultaneously by shot quality, match state, field setting and pitch condition; football's shot mostly hinges on location and body part. My translation is ERV — Expected Run Value: the runs expected from a ball in a specific situation. This is not xG; it is xG's Bengali cousin. Where the mapping breaks: in football a shot always creates surplus opportunity, but in cricket a dot ball carries negative ERV — a debit. That negative number is the centre of this entire analysis. Why hand-log at all? Because analysing Bangladesh requires building the data before analysing it. Ball-tracking feeds, ground-truth labels, venue-adjusted scorecards — these luxuries are scarce in South Asia's domestic ecosystem. The analytical culture born here is shaped less by talent than by scarcity. I will not hide my sample: fourteen T20 international chases by Bangladesh between January 2026 and December 2026, every one hand-coded ball by ball, all format-specific, all chase-specific, with no first-innings data mixed in. Fourteen innings is a small sample. Leaping from a small sample to a large conclusion is the oldest sin in data analysis, and I refuse it. Read this as an audit trail, not a verdict. Now the core. I built three proprietary indices from the ball-by-ball record. First index: Dot-Ball Sequence (DBS) — how many consecutive dot balls fall, and in which overs. Across the fourteen chases, 58 percent of Bangladesh's total dot balls fell between overs 7 and 15. Only 29 percent fell in overs 16 to 20. Globally, dot balls cluster most heavily in the powerplay, where field restrictions and a new ball apply. Bangladesh's distribution is inverted — the middle phase is the driest. That single number relocates the entire debate. The death-over 'collapse' is not a collapse; it is the final settlement of pressure accumulated long before. Second index: Required-Rate Elasticity — how much the required rate rises per dot ball. Across the fourteen chases, before the 15th over each dot ball added 0.31 to the required rate on average. After the 15th over, the same dot ball added 0.47. Same act, different price. A dot ball in the 12th over means time remains. The same dot ball in the 17th means time is gone. This is why batters attempt abnormal shots at the death — what the model calls a constrained decision. Calling it mental fragility is possible, but that turns the eye into judge rather than witness. The model says the pressure was banked earlier; the death overs merely pay the interest. Third index: Death-Over Entropy — not chaos of shot-making but uncertainty of outcome. In the last five overs, every ball can produce four outcomes: dot, one or two runs, boundary, wicket. Entropy measures how evenly outcomes spread. High entropy means unpredictable; low entropy means near-determined. In Bangladesh's fourteen chases, entropy fell in the final five overs because outcomes clustered at two poles — dot or wicket. The middle band of 'one or two runs' nearly vanished. Elite chasing sides show a wider middle band, because they rotate strike under pressure. One clarification. I have worked on pressing, having analysed Italy's Euro 2026 campaign — a PPDA of 7.2 across seven matches, the tournament's lowest. The lesson from that does not transplant directly, because cricket pressing is ball-based, not team-based. But the principle holds: pressure is not a mood, it is a ledger. Death-over bowling is not chaos; it is an account book. A bowler who knows what one dot ball in the 17th over does to the opponent's required rate is not watching the crowd — he is watching the batter's wrists. Before player level, a venue correction is mandatory. Mirpur's Sher-e-Bangla surface behaves differently — slower, more reliable, spin-friendly. Chasing 170 there is not the same task as chasing 170 in Chattogram or Sylhet. Seven of my fourteen innings were at Mirpur. There, the average starting required rate per over was 8.1; in the other seven, 8.9. That gap proves that without venue adjustment, Bangladesh's 'weak chasing' statistic is half a truth. Where runs are hard to score, a dot ball costs less; where runs flow, a dot ball is poison. Now the named accounts. Towhid Hridoy's case is the most instructive. His middle-over strike rate is stable in our log, but his mode of contact differs — he plays the gaps, and those shots yield twos rather than fours on slow pitches. A two is half a four, and it consumes a ball. Litton Das is the inverse: he sets the tempo in the powerplay, but between overs 9 and 14 his boundary distance shrinks and dot balls begin to accumulate. Mahmudullah's number is the most uncomfortable — across the fourteen chases his strike rate after the 15th over is the highest in the log, but the balls remaining in front of him are the fewest. He does the job, very late; in the model's language that is not skill but a timing deficit. Shakib Al Hasan raises a different question. His strike rate reads conservative, but his wicket risk per ball is the lowest in our log. Chasing requires reading those two together — risk and output. A low-risk batter who consumes balls in the middle overs costs the team, yet the individual scorecard shows 'not out 30'. Team accounting and individual accounting diverge here, and television usually shows the individual one. From the bowling side, the opposition plan is no longer mysterious. Pairing slow bowlers with leg-spin between overs 9 and 14 generates pressure against Bangladesh, because that is where Bangladesh's boundary risk is lowest. Afghanistan's Rashid Khan and Naveen-ul-Haq in the 2026 World Cup Super Eight produced a textbook rendering of that design; the batters did not play bad shots — they simply could not identify which ball to attack. And precisely here the model dismantles a comfortable myth: teams that lose at the death did not actually lose at the death. Now the section where I must re-audit my own reading — the eye test, in a bounded role. From both the commentary box and the dugout, the scene I observe is Bangladesh's body language shifting as they enter the 16th over — shoulders drop, calling slows, run-out risk rises. That testimony is probably true, but it is not proof; it is a hypothesis generator. In my log that hypothesis held in only seven matches. In the other seven, the chase had already become mathematically near-impossible by overs 10 to 12, and nobody looked panicked. When the eye and the model disagree, I do not rule — I publish the disagreement. An analyst who makes the eye the judge finds stories easily; one who keeps the eye as witness finds truth slowly. Here is the real disagreement, and it inverts my own central argument. The conventional reading is that Bangladesh collapse at the death and therefore need hard-hitters. The numbers say otherwise: in ten of the fourteen chases, the required rate had already crossed 12 before the 15th over. By the time the death overs arrived, win probability was already under 20 percent. Where is the death-over failure in that? In a match with an eight percent mathematical chance, losing is not failure — it is outcome. We hunt for failure where none exists, while the actual leak runs silently through overs 7 to 12. Second disagreement: correlation versus causation. The claim that more hard-hitters improve death overs finds statistical support, because good teams have both. But a third variable sits in between — pitch character and middle-over strike rotation. Strong rotation reduces the need for hard-hitters; abundant hard-hitters without rotation pushes everyone to swing beyond the boundary in the last five overs, collapsing entropy to two poles, exactly as observed. You can carry a 2026 statistic into a 2026 decision only if you declare format, venue and opposition bowling depth alongside it. Third disagreement: natural experiments. In 2026, home advantage fell in empty football stadiums — home wins from 43.2 to 33.7 percent, goals from 3.1 to 2.7. Assuming that transfers literally to cricket is a guess; cricket's home advantage leans heavily on pitch preparation and travel, less on crowd noise. So I will not play the 'ghost stadium' card. One thread is relevant though: if pressure arrives from the external environment, then pressure-management training should be durable. For Bangladesh the numbers suggest the opposite — the training ledger barely exists. So what do I watch in the next series? I pre-register the thresholds now, so no story can be manufactured later. In the next chasing series at Mirpur I will track three things: if dot-ball percentage in overs 7 to 12 crosses 35, that signals a model recalibration; if non-boundary strike rotation in that same phase sits below 45 percent, the middle order needs restructuring; and if the required rate at the 15th over is under six, every death-over debate is irrelevant — that is an easy chase, where failure means a decision failure, not a capability one. A model is a monastery. You enter with noise, and you leave with discipline. My bedroom spreadsheet still opens every match, and it still reminds me of one uncomfortable thing: Bangladesh's chase does not die in the 16th over. It dies in the 11th, quietly, in an unrecorded entry of a dot ball. When the next chase begins, watch the first ball not for the shot — watch the strike rotation.