When Run-Expectancy Went Silent: An Autopsy of Bangladesh's Batting Model at the T20 World Cup

At the Grand Prairie Stadium in Dallas that evening, two numbers sat side by side on my laptop screen. One was the real score; the other was my run-expectancy model's projected score. In the seventeenth over, Bangladesh had 112 on the board with eight overs left. The model said that from that position, a top-eight international T20 side reaches roughly 162. Bangladesh finished on 147. A fifteen-run gap — in T20, that is usually the line between winning and losing.

But the real earthquake that evening was not on the scoreboard. It was inside my model. Across the first four matches of the tournament, the same model had tracked Bangladesh's batting output to within four or five runs. In the fifth match, it suddenly went silent. And the place where it went silent was exactly the place — the middle overs — where Bangladesh's T20 fate has been written, year after year.

In Rajshahi, the run-expectancy column stopped being a number and became a confession.

Context: Why a Model Has to Be Translated

I have worked with football's xG model since 2026, and since the 2026 World Cup in Russia I have tried to translate it into cricket's discrete-event world. In football, xG tells you how promising a shot was — whether the goal actually went in is a separate question. In cricket, its closest relative is run expectancy: how many runs a side averages from a given over, wicket and situation. Both are languages of probability. Both pull the outcome off the table of fate and onto a table where it can be measured.

Before the 2026 T20 World Cup began, I built a run-expectancy baseline for Bangladesh. There were four inputs: powerplay run rate, boundary frequency between overs seven and fifteen, death-over strike rate, and the rate at which wickets fell per innings. The outputs were a projected score and a win probability. I did not use it to watch the game; I used it to catch the game's miscalculations.

For the first two weeks of the tournament, the model was almost flawless. Bangladesh beat Sri Lanka, beat the Netherlands, beat Nepal. The model had already said that Bangladesh's win probability in those three matches would sit above fifty per cent, but that the scorelines would never be enormous. That is exactly what happened. Narrow wins, narrow fear. Then came a four-run defeat to South Africa, and in the Super Eight the model suddenly turned to stone.

I understood then that the problem was not in my inputs. The problem was in a truth of cricket that a football model never teaches you: in cricket, batting is not only a machine for making runs; it is simultaneously a machine for managing risk. And in Bangladesh's case, the price of that risk is usually paid back, with interest, in the middle overs.

The Core Analysis: Where the Numbers Stop

Split Bangladesh's tournament batting data into three phases and the picture sharpens. In the first six overs, Bangladesh's run rate was about 7.8 per over, against a top-four average in the low 8.4s. That gap is not dangerous; falling half a run behind over six overs does not lose a side a match.

The real collapse came between overs seven and fifteen. In those nine overs, Bangladesh's strike rate in my model sat between 105 and 110. Over the same stretch, India, Australia and England posted middle-over strike rates of 130 to 140. That means Bangladesh were falling roughly two to three runs behind every over — which compounds into twenty to twenty-seven runs across nine overs. In T20, twenty runs is about one over of breathing room.

Where does that deficit come from? Tactically, from two places.

First, Bangladesh's middle-over batting is governed by an anchor culture. From overs seven to fifteen, their primary job becomes protecting wickets so they can attack in the last five. The problem is that in modern T20, the right to attack in the last five overs is bought in the middle overs. A side batting at a strike rate of 105 in the ninth over walks into the last five overs under the most favourable conditions for the opposition's pace bowlers — because the ball is older, the boundaries feel shorter, and the bowling side knows the batters are now forced to take risks.

Second, Bangladesh's boundary frequency in the middle overs is abnormally low. The tournament's leading sides were hitting 1.1 to 1.3 boundaries per over between overs seven and fifteen. Bangladesh's figure was below 0.7. When boundaries do not come, runs arrive one and two at a time, and the opposition controls the strike rate.

This is where my model was wrong — and the error is the most instructive part. My run-expectancy model treated Bangladesh's middle-over strike rate as a stable variable, based on the six months of data before the tournament. But in the Super Eight, the standard of the opposition changes. South Africa, India, Australia — their spin and pace are both a step above Bangladesh's. The model did not give enough weight to the variable called quality of opposition. That was my modelling error, not the players'.

I stopped watching boundaries and started reading the empty spaces before them. What I found there was a structural picture: Bangladesh's batters were leaving more balls in the middle overs and hitting fewer. In football's language, they were not taking shots — they were only playing passes. And a side that does not shoot does not raise its xG, just as in cricket a side that does not hit boundaries does not raise its run expectancy.

A model graph makes the point easily. A typical Bangladesh innings produces a nearly flat run curve — a steady two to six runs an over, but no over in which seven or eight runs explode. The leading sides' curves are jagged: a few quiet overs, then one explosive over that drags the average upward. In T20, winning means a jagged curve; a flat curve means control — but control never writes a win on the scoreboard.

The death-over data clarifies this structure further. Bangladesh's strike rate in the last five overs was around 150, which is not poor — but the leading sides posted 175 to 190. The reason for that gap is again hidden in the middle overs. A side that reaches fifteen overs at 95 for 4 does not have two set batters for the last five; it has new batters and added pressure. So the death-over strike rate falls — because the number of set batters falls.

The number of boundary-less overs Bangladesh played in the middle overs was abnormally high. In some innings, three or four consecutive overs passed without a single four or six. In international T20, a run of these dry overs means the opposition's bowlers get time to experiment with their lengths — and if the experiment works, it is used again in the next match.

One more statistic completes the picture. The balls-per-boundary figure for the batter who came in at number three was about 14. For the leading sides, that figure was 8 to 10. In other words, the batter occupying Bangladesh's most important position was hitting a boundary once every fourteen balls, while rivals were doing so every eight to ten. Number three is T20's tempo setter — the player who determines the innings' speed. If that position is slow, the whole innings is slow, because the batters who follow inherit that tempo.

One point needs to be made clear, because there is much confusion around it. Bangladesh's batters do not lack talent. Towhid Hridoy, Litton Das, Najmul Hossain Shanto — all are of international standard. The problem is not talent; the problem is role — the decision about who does what in which phase. In T20, roles are set before the match, not within it. And in Bangladesh's role-setting, an old formula still operates: look first, then hit. Modern T20's formula is the reverse: hit, then see how much you can manage.

Against Afghanistan in the Super Eight, this structural weakness was exposed most ruthlessly. The Afghan spinners bowled immaculate lengths in the middle overs, and Bangladesh's batters lost wickets searching for boundaries. Here is a curious paradox: when Bangladesh tried to attack, they lost wickets; when they stayed defensive, the run rate sank. The skill of building a bridge between those two is modern T20's real currency — and it is what Bangladesh have least of.

Cross-Sport Translation: Pressing and the Middle Overs

In 2026 I covered the Euros and the Tokyo Olympics at the same time. In the Euro final, Italy's 1.7 xG against England's 0.9 told the story — Italy controlled the match; the penalty shootout was merely the language of the result. In Tokyo, Elaine Thompson-Herah ran 10.61 in the 100m and 21.53 in the 200m. I began to draw a parallel between pressing intensity and sprint recovery: both are an accounting of how much pressure you can absorb and how much you get back.

In cricket that parallel is even more direct. Just as football's pressing creates pressure in a specific zone and forces the opponent into error, attacking in T20's middle overs creates pressure in a specific phase and buys an advantage in the next. Bangladesh do not make that purchase. They save in the middle overs, but the savings are not returned with interest in the final overs — because the final overs are the most expensive.

In 2026 I worked on empty stadiums. I saw then how boundary size, travel and rest days shape results. At the 2026 T20 World Cup, venue and travel were a hidden variable too: between the United States and the West Indies, teams had to hop from one country to another on short flights. For Bangladesh, that travel load compounded the middle-over slowdown. But I will be honest — I could not properly measure the effect of that travel load; my data lacked that granularity. A variable you cannot measure, you can only suspect, not prove.

Value Note: The Ground the Market Considers Safe

A value note is warranted here, because cricket is now not only a game on the field but a game in the market. Franchise T20 leagues price batters largely on one index: powerplay and death-over strike rate. They weight middle-over strike rate far less, because many models assume the middle overs are safe — that nobody loses there and nobody wins there. Bangladesh's middle-over batters are priced low precisely for this reason. Yet the tournament's data says the middle overs are the most valuable of all — because that is where a match's fate is decided. The ground the market considers safe is in fact the most dangerous.

Contrarian Angle: Not Mental, Structural

Here is where I object to a conventional wisdom. A comfortable story has long circulated in Bangladesh's cricket analysis: our batters collapse mentally on the big stage. I reject that story, because the data says otherwise.

If the problem were mental pressure, the pressure should be greatest in the match where winning matters most — the last group-stage game. But Bangladesh's group-stage performance was the most stable part of their tournament. The collapse came in the Super Eight, where the opposition was stronger — meaning the problem is not mental, it is structural.

Rather, the reverse is true: the way Bangladesh bat in the middle overs is a fear-based strategy — but the fear is not mental fear, it is accounting fear. Fear of losing wickets. In modern T20, a wicket is an asset, but it is an asset to spend, not to hoard. Bangladesh hoard wickets; they do not spend them. So at the end of the innings, a great deal of the asset remains in hand — and far too few runs to win.

One more point. Many say Bangladesh's bowling is good and their batting weak. The data does not respect that division either. Bangladesh's bowling is as skilled in the middle overs as their batting should have been in a different phase. The problem is not batting versus bowling; the problem is phase versus phase. They are doing the right things at the wrong times.

Takeaway: The Signal for the Next Innings

The World Cup did not create value; it simply turned the lights on — and in that light it became visible that Bangladesh's batting-model problem is really an invisible leak in the middle overs. For the next tournament I have already made one decision: in my run-expectancy model, I will no longer treat quality of opposition as a fixed constant. It will become a moving index, recalibrated before every match.

The signal is patient; the noise is always in a hurry. In Bangladesh's cricket, the noise always surrounds individual performance — a century, a five-wicket haul, a catch. But the signal sits quietly, between overs seven and fifteen, where nobody looks. The next time Bangladesh bat through the middle overs, my eyes will not be on the scoreboard — they will be on the strike rate across those nine overs. Because that is where, in the place where the numbers go silent, the true result of the match is already written.