From 108/2 to 149: A Ball-by-Ball Forensic of Bangladesh's Collapse

One number on the scorecard was telling the truth: at 12.4 overs Bangladesh were 108 for 2. What followed were 40 balls, 41 runs, 8 wickets. A 19-run defeat to Sri Lanka in the Asia Cup 2026 Super Four— but the defeat isn't written on the scoreline, it's written inside the dot balls of those last 40 deliveries. After the match I hand-charted all 116 deliveries of the innings, because television graphics show you where the loss happened, not why. One number kept returning: in the last 7.2 overs Bangladesh played 23 dot balls— roughly one runless ball every second delivery. In T20 that is no longer a batting failure; it is a structural fracture. Before you trust a number you have to know its environment— pitch, dew, toss, workload. I learned after hand-logging 9,714 shots that a number without context is meaningless. At the R. Premadasa Stadium in Colombo, an evening match means dew is a familiar variable; in the second innings the ball is hard to grip, spinners' lines shorten, and slower deliveries skid onto the bat. Sri Lanka, losing the toss and batting first, made 167 for 6— the first warning, because they scored 58 in the last five overs while Bangladesh's seamers kept missing the yorker length. It also matters that this was Bangladesh's third match in five days— the workload of two frontline quicks was already at the red line. Taskin Ahmed had bowled 8 overs across the previous two games, Mustafizur Rahman 7.4. The men they needed at the death were already tired. That fatigue doesn't lose a match directly, but it nudges length by an inch or two— and in T20 two inches of length is six runs. Before Bangladesh's innings the equation was simple: 8.35 an over, wickets in hand. Litton Das and Tanzid Hasan built a base in small partnerships, but there was no fluency. The powerplay produced 48 for 1— with 18 dot balls. The start was built on run-pressure, not flow. The first signal was hiding there, and nobody read it. I split the innings into three phases— powerplay (1–6), middle (7–15), death (16–20). The numbers talk. The powerplay yielded 48 for 1 in six overs, with 18 dot balls from 36— 50%. Najmul Hossain Shanto, coming in at one down, started slowly, clearly trying to set himself. The middle overs, 7 to 12.4, produced 60 runs with 11 dot balls from 40. Up to 12.4 the partnership was clean: 8.5 an over, reasonable strike rotation, and two sixes off Towhid Hridoy's bat gave the innings momentum. 108 for 2— from there my model put the win probability at 62%. What followed was not merely 'wickets fell.' Wanindu Hasaranga came on in the 13th over and changed his length— flatter, skiddier, slightly shorter. Bangladesh's batters went searching for the cover drive and stopped rotating strike. Overs 13 to 15: 14 runs, 2 wickets, 10 dot balls from 18— 56%. The required rate jumped from 8.18 to 11.60. That is where the match turned, and it was turned not by a wicket but by a dot ball. At the death, Maheesh Theekshana and Dushmantha Chameera bowled a mix of yorkers and slower bouncers, and Bangladesh's lower middle— Mehidy Hasan Miraz, Taskin Ahmed— could not survive. The last 4.2 overs: 19 runs, 4 wickets. Across the innings there were 54 dot balls from 116, or 46.5%, nine points above the tournament average. The biggest number is this: of Bangladesh's 149, 88 runs came from 16 boundaries— 59%. Nearly three-fifths of the runs came from outside the rope, and almost nothing came from the ones and twos inside it. There was no strike rotation; there was only big-hit dependence. When boundaries stopped— Hasaranga's googly, Pathirana's yorker— there was no alternative route to runs. Wickets fell in a cluster because every ball was a boundary attempt. I tracked one thing separately: after the 13th over, Bangladesh's batters took just 3 singles on the off side from 21 deliveries. The field was set, and Bangladesh did not change the plan against it. That is not a failure of batting technique; it is a failure of innings management— some will call it luck, but 3 singles from 21 balls is not luck, it is a pattern. The easy explanation is 'the lower order is weak.' But that mistakes correlation for causation. Suppose Bangladesh had taken a single an over from the 13th and made 60 off 45— they would have won, and the same batters would not have been labelled weak. So where is the real fault? First, in team composition. Bangladesh picked a pure hitter at seven but no strike-rotator. Second, their death-overs plan was 'boundary or nothing'— there was no realistic two-run plan. Third, the toss and dew. When I hand-logged 92 matches of empty-stadium football in 2026, I learned that a crowd is not noise, it is a variable; every empty stadium rewrote a coefficient I thought was stable. Here too the environment was shifting— the pitch was slowing, the spinners were gripping the ball— yet Bangladesh held the old tempo instead of adjusting to the change. My model said the win probability from 108 for 2 was 62%, but that number was not wrong— it was incomplete. Because the model did not know how slow the pitch would get after the tenth over, or how strike-rotation-incapable Bangladesh's lower order was. A model without its environment is just a rumour with decimals. For the next match Bangladesh face one question: is strike rotation the priority, or big-hitting? In this Asia Cup format, on this dewy, slow pitch, the skill of reducing dot balls will win matches, not the number of sixes. If anyone says 'it was only a batting failure,' they have not looked at the 54 dot balls. Before the next Super Four game, hold one number in mind: 46.5%. Until it drops below 40, this collapse will keep returning.