Overs 7 to 15: Where Bangladesh's T20 Batting Actually Loses Its Matches

A single innings keeps returning in my coding sheets. Eleven overs gone, the score 68/2, the set batter 34 off 29. The next twenty-four balls produce 19 runs: one four, the rest singles and dots. Nobody is dismissed. No wide is bowled. The scorecard will say the side was "steady." The scorecard does not lie, but it does not tell context either. Across those twenty-four balls the deliveries landed slow, the field spread deep, and Bangladesh's batting drifted under six an over in a tournament where eight of ten sides were above seven.

That drift is the subject here. We have mourned the powerplay for a decade, yet the 68 T20 matches I have hand-coded say Bangladesh's batting is actually lost between overs 7 and 15. No wickets fall there, no runs come either, and the broadcast camera is busy looking at the physio.

Context: why the count mattered

Every tournament cycle, the conversation spins in the same place. Flags, stories, nostalgia. Before the first match, everyone says the same sentence: if our powerplay is fixed, everything is fixed. From 2026 through 2026, almost every Bengali-language preview repeated that line. Nobody once asked what happens after the powerplay is fixed.

I have coded matches since 2026. It began with 22 matches, 1,140 possession sequences, forty variables per sequence. What I learned there was simple: patterns hide in the empty spaces, not the highlight packages. In T20, the empty space is overs 7 to 15.

Let me state the method plainly, because I do not publish a percentage without its denominator. This analysis uses 68 matches: 27 BPL matches across the 2026, 2026 and 2026 seasons, plus 41 Bangladesh international T20s between 2026 and 2026. That is 15,912 legal deliveries, each coded for phase, batter, bowler type, shot direction, field setting and match state. Where my hand count diverged from the official feed, I kept my number and flagged the gap separately. It diverged in four matches, all rain-affected.

One limitation up front. Six venues, six different pitch characters, and innings samples that sometimes run to only six balls. I used a twelve-ball window because a twenty-four-ball window led me to a wrong conclusion last year. A bigger sample is not a truer one; an adequate one is.

The full picture: splitting the innings into phases

Powerplay (1-6): Bangladesh run at 7.84, roughly at the global mean, 0.39 below. Death overs (16-20): 9.62, 0.21 below the mean. Both ends of the innings, the side keeps pace.

Middle overs: 7.02. Against six comparable sides I selected on squad depth, the average is 8.31. That is a gap of 1.29 runs per over, which across nine overs is 11.6 runs. Of Bangladesh's six defeats inside ten runs since 2026, five showed that same middle-phase shortfall.

My core reading: roughly sixty percent of the runs the side loses across the powerplay and death overs combined is actually accumulated, or rather forfeited, between overs 7 and 15.

This is not a question of which phase is good or bad. It is a question of where runs accumulate and where they stall. Powerplay weakness is visible because the field is up and the commentary is busy. Middle-over weakness is invisible because everything looks normal.

The quiet arithmetic of dot balls

Bangladesh's dot-ball rate in the middle overs is 38.6 percent. The six comparison sides sit at 31.2. That 7.4-point gap is roughly 45 to 50 wasted balls per innings, in the middle phase alone. Boundary rate: 11.4 against 14.8. Singles rates are nearly identical, both around 30 percent.

So the problem is not the ability to take runs, it is the wastage of balls in the windows where no boundary arrives. Small dots pile up and leave the side, in the last five overs, in a position where even two clean sixes are not enough. I counted twenty-two matches by hand; the spreadsheet remembers what injury and poor record-keeping erased. That forty-ball gap is where the real loss sits.

The twelve-ball cluster: an old metric in a new phase

In 2026, my 22-match dataset surfaced a pattern I could do nothing with. Sixty-eight percent of goals conceded came within twelve minutes of losing the ball in their own third. I translated that football metric into T20 as the frequency of two wickets falling within twelve balls.

Bangladesh have 21 innings where they were bowled out for or finished under 120. In 14 of them, two wickets fell within twelve balls. Nine of those pairs fell between overs 7 and 15. Not one fell in the last two overs.

This is my most-used metric because it speaks the language of bowler and batter decision alike. When a set batter cannot score in the middle overs, one of two things usually happens: he plays a wrong shot to lift his strike rate, or a new batter at the other end deepens the pressure. Both recur in my data.

Injury-adjusted records: how much of the 68 was a full side

This is where the work becomes hard, and where I spend the most time. In 31 of the 68 matches, at least one first-choice top-six batter was absent. Injury, rest, visa issues, concussion protocol. The XI on the field was not Bangladesh's best XI.

I isolated 11 matches by hand in which, by my count, the batting line-up was near full strength. In those 11, middle-over run rate was 8.14, strike rate around 130. Across the other 31, run rate was 6.71. The sample is small, one match shifts the average by several points, so I make no final claim. Still, the signal points one way: a large share of the shortfall is about selection and fitness, not skill.

Here, the empty-stadium study taught me to wait until the evidence closes. 1,200 matches, 412 behind closed doors, home win rate falling from 44.8 to 37.6 percent, home penalties down 19 percent. I wrote no "new normal" prediction until that sample closed. Same discipline here: 11 matches will not carry a conclusion, but they will fix what I watch next.

The structure problem at No. 5

Across forty internationals, Bangladesh's No. 5 batters faced 711 balls at a strike rate of 118.4. Over the same period, the No. 5s of sides reaching tournament semi-finals averaged 138.7. The gap is twenty, more than two balls an over.

The selection philosophy behind it is clear. In most matches, five and six were occupied by all-rounders who also bowl. As balance, that is defensible. But if that balance does not buy batting strike rate, then it is really locking away half the side. Where a specialist batter has appeared at seven, the team has scored about six runs more on average.

I am not arguing that all-rounders are bad. I am arguing that carrying weakness in two slots at once shows up in a third. In T20 it shows up in the middle overs, when the bowling side blocks two dots and identifies exactly whose strike rate is under pressure.

I have watched this from the ground, where television does not reach. In the middle overs, fielders drop two steps, long-on and deep midwicket drift toward the rope, and a No. 5 whose first stroke is the pull or the sweep cannot lift the ball into that space. Those two steps are invisible in the data. They live in scouting notes.

The contrarian angle: correlation is not cause

Now the honest work, honestly. A good middle-over run rate correlates with a good start; it does not follow from it. A side whose top order absorbs pressure leaves less pressure on five and six, so their strike rate looks better by default. My analysis has that hole.

I tried to fill it by discarding powerplay-driven matches. In the 29 matches where the powerplay score sat within ten runs of par either way, the comparison is fairer. Even in that subset, the middle-over run rate gap is 0.94, smaller, same direction. It does not prove the idea. It keeps it alive.

Another thing must be said that few want to write. Criticising the powerplay is safe. You can demand a coach's removal through the powerplay, the number looks sharp in a report, and the board agrees easily. Fixing the middle overs requires changing selection philosophy, which is riskier, which invites deeper questions. A coach switching from a back four to a back three is making the same calculation: he is avoiding reputational risk, not solving a problem. Selectors put another anchor at No. 4 for exactly the same reason.

The Croatia piece was right; the market simply did not listen on time. That night the model had 14 goals against 8.9 xG across seven matches, with three knockout wins built on two shootouts and an extra-time goal. The narrative buried the model. I have the same fear about the middle overs narrative.

And the pitch, which changes from ground to ground

On slow, turning surfaces, Bangladesh's middle-over dot-ball rate is 42.1 percent; away from home it is 35.9. A 6.2-point gap. That is an easy explanation: on slow pitches boundaries are harder, so strike rate falls. True, but partial. The opposition bats on the same surface, and their dot-ball rate is 34.7. Nearly seven points of difference survive.

Of my 68 matches, 34 were on slow-to-medium surfaces, 22 on good batting decks, 12 on rain-dampened pitches. The middle-over shortfall holds across all three. No single pitch character can be its sole cause.

What I will watch in the next series

I trust no narrative until I count it myself. So I register a test in advance, with a date, so nobody can later say I changed my explanation.

First signal: who bats at No. 5. If a bowling all-rounder occupies it again and the strike rate stays under 125, my arithmetic holds. Second signal: the rate of two wickets falling within twelve balls between overs 7 and 15. If it stays above 65 percent, the problem is selection, not planning. Third signal: middle-over boundary rate. Clear 13 and a large part of the deficit disappears inside one series, through a single small decision.

If you write percentages, you write denominators. That was my first discipline and it is also my biggest constraint. Six balls of middle-over sample per match will never carry a final verdict. But 68 matches, 6,084 balls, 21 sub-120 innings, and one clear direction will. If Bangladesh fix the powerplay again next tournament and stall in the middle overs again, the question will no longer be about the batters. It will be about who chose, why, and with how much interest, the four slots from three to six. The decision no sample captures is the one that loses the most matches.