The Dot-Ball Ledger: A Ten-Match Audit of Bangladesh's T20 Middle Overs

MIRPUR, SECOND T20I OF THE LAST HOME SERIES. At the ten-over mark the board read 78 for 2, chasing 166. In the commentary box the talk was about overs 16 to 20 — who takes the big shot, who holds strike, whose two overs get saved. The match ended 19 runs short, and the blame landed on two failed big shots in the 18th and 19th overs.

My scorebook was pointing somewhere else. In that innings, overs 10 to 15 produced 21 dot balls out of 36. Fifty-eight percent of the deliveries in that six-over stretch produced nothing. In the last five overs there were five dot balls. The phase no camera watches, the phase that never raises a fan's pulse — that is where the game turned.

I have watched this same scene for years. It is exactly why, in 2026, I put a rule on myself: no tactical claim without ten matches of data. The Burnley thread that year looked like noise until I sorted it by phase baseline — then the story was plain in the numbers. Cricket teaches the same lesson with different units. Football's pressing metric and cricket's dot-ball metric are not the same thing, but the question is identical: where did the match turn, and why?

CONTEXT: WHY PHASES, WHY BASELINES, WHY TEN

I split a T20 innings into three phases. Powerplay, overs 1 to 6. Middle, overs 7 to 15. Death, overs 16 to 20. Those boundaries are not arbitrary. Delivery type, field setting, bowler responsibility and batter risk appetite all change at those seams. Judging a batter by one aggregate strike rate means blending three different games into a single number.

My definitions, published so readers can audit them: strike rate is runs per hundred balls — simple, but blind to dismissal risk. Control percentage measures whether the shot played was the shot intended, separating a middled ball from an edge or a mis-sweep. Dot-ball percentage is the share of deliveries in a phase that yield nothing — the least discussed and most damaging number in the format. Boundary dependency is the share of a phase's runs that came in fours and sixes. Phase-adjusted run rate is what remains after weighting era, venue, opponent and pitch type. Without that adjustment, a 120 in 2026 and a 120 in 2026 look identical. They are not. I have walked into wrong conclusions by skipping that step, and I do not intend to repeat it.

Why ten matches? Because a middle-order batter faces roughly 15 to 20 balls per innings in the middle phase. Ten matches is 150 to 200 balls, and phase strike rate stabilises reasonably well at that sample. But the threshold must be declared before the result, never fitted to it. If I move it, I write down why — on spin-dominant surfaces the middle-phase sample stabilises faster, so six matches can be defensible there.

Data source and cleaning: ball-by-ball logs from the last twelve T20Is, hand-tagged for seam or spin, wide, and dead ball. Powerplay overs with the extra fielder are flagged separately. Wides and no-balls are not counted as dots. One further note, relevant to how seriously this work should be taken: if ball-by-ball logs sat on an immutable ledger, the question of who quietly edited a statistic would have an answer. Cricket boards have not gone there yet. For any analyst's evidence file, immutability is a moral inevitability, not a novelty.

CORE: THE EVIDENCE CHAIN

Start with the whole picture. Bangladesh's phase strike rates over the last twelve T20Is against a comparable world baseline of the top eight sides, similarly condition-weighted:

| Phase | Bangladesh SR | World baseline SR | Gap | |-----|-----|-----|-----| | Powerplay (1–6) | 128.4 | 133.7 | −5.3 | | Middle (7–15) | 108.9 | 128.5 | −19.6 | | Death (16–20) | 156.2 | 158.0 | −1.8 |

The gap lives in one phase. Five runs behind in the powerplay, effectively level at the death, nearly twenty behind in the middle. That is nine to ten runs per innings — decisively inside the margin between winning and losing this format.

Now the dot-ball ledger.

| Phase | Bangladesh DOT% | World baseline DOT% | |-----|-----|-----| | Powerplay | 46.1 | 45.2 | | Middle | 44.0 | 32.8 | | Death | 31.4 | 30.1 |

Powerplay dot rate sits on the baseline, because everyone eats dots against the new ball. The problem starts at over seven. The baseline drops to 33 percent in that phase; Bangladesh stays stuck at 44. Four empty balls in every nine. That is not rhythm. That is stasis.

Not all dots are equal, though. Break them down by type in the middle phase: 38 percent defensive blocks, 34 percent missed sweeps or broken shots, 28 percent leaves or singles turned down. A high share of the second category tells you the issue is method, not intent.

| Bangladesh, middle phase | Share of dots | |-----|-----| | Defence / block | 38% | | Missed sweep, broken shot | 34% | | Leave or declined single | 28% |

Then boundary dependency: 41 percent of Bangladesh's middle-phase runs came in fours and sixes, against a baseline of 52 percent. That table reads backwards at first. Low boundary dependency sounds like low risk. But in the middle overs it can mean one of two things: either you are rotating strike well, or you are neither rotating nor hitting boundaries while the required rate climbs. The proof against the first reading is the strike rate itself: 108.9. Genuine rotation produces close to 125.

WHERE THE BATTERS GET STUCK

Ten-match window, middle phase only, minimum 80 balls faced.

| Batter profile | SR (7–15) | Control % | DOT% | |-----|-----|-----|-----| | Right-hand opener | 114.2 | 79.1 | 41.6 | | Left-hand opener | 106.7 | 76.4 | 45.2 | | Number three | 111.9 | 78.0 | 42.8 | | Middle-order specialist | 129.4 | 82.6 | 34.1 | | Finisher (batting lower) | 119.8 | 74.9 | 38.7 |

The problem is not individual weakness. It is infection. Even the middle-order specialist with the best control percentage sits at 34 percent dots in that phase — level with the baseline, but with one in three run balls coming from a six attempt. The other four roles sit below baseline.

I then pushed each innings through a rolling ten-match window.

| Window | Middle-phase SR | Gap to baseline | |-----|-----|-----| | 1 | 103.1 | −25.4 | | 2 | 107.8 | −20.7 | | 3 | 112.4 | −16.1 | | 4 | 109.2 | −19.3 | | 5 | 114.6 | −13.9 | | 6 | 106.3 | −22.2 | | 7 | 115.8 | −12.7 | | 8 | 110.9 | −17.6 |

Seven of eight windows show a double-digit deficit. One window breaks away — the seventh, improving to −12.7. Where did it come from? Three home series in which two of the opposition attacks were built around off-spin. Had it been a trend, window eight would have held. It fell back.

THE PRECEDENT TABLE: NO COMPARISON WITHOUT ERA ADJUSTMENT

T20 scoring is inflating. A precedent table without era weighting teaches the wrong lesson.

| Era | Global middle-phase SR | Bangladesh | Adjusted gap | |-----|-----|-----|-----| | 2026–17 | 118.9 | 112.0 | −6.9 | | 2026–21 | 123.4 | 110.5 | −12.9 | | 2026–23 | 126.1 | 107.2 | −18.9 | | 2026–26 | 128.5 | 108.9 | −19.6 |

This is the most uncomfortable table in the file. Over a decade the world baseline climbed 9.6 runs. Bangladesh climbed 3.1. The world accelerated; Bangladesh kept walking at the same pace. This is not one series. It is a ten-year slope.

The bowling side matters here, because it is being buried.

| Phase | Bangladesh economy | Baseline economy | |-----|-----|-----| | Powerplay | 7.42 | 7.80 | | Middle (spin overs) | 7.94 | 8.24 | | Death | 8.87 | 9.31 |

And the stability check across bowler types — eight matches, tolerance of one run per over: left-arm seam 9.10 death economy with 36.4 percent dots; right-arm new-ball seam 8.32 with 41.2; leg-spin 7.68 with 38.9; off-spin 7.91 with 42.1. All four hold inside tolerance. The bowling method is not the problem in this cycle. The middle-phase batting method is.

THE ANATOMY OF THE INFECTION

Four linked causes. First, the consolidation mantra in overs 7 to 11. In my tagged logs, strike rate in the five overs after a wicket falls drops to 94; the baseline in that match state drops only to 119. Risk aversion is trained into the side.

Second, no designated spin-hitter. Against spin in overs 7 to 15, Bangladesh produce 0.49 off-side boundaries per over against a baseline of 0.81. Against leg-spin the wide long-on region is almost never used. One player can sweep; three cannot. Dependence concentrates, and the opposition sets its fields accordingly.

Third, strike rotation. Singles percentage in the middle five overs is 31.4 against a baseline of 38.9. In T20, the single is the silent engine. Where it is missing, one extra high-risk shot every two overs becomes compulsory.

Fourth, the set batter's call. In the ten balls after a new batter arrives past the tenth over, the run rate is 4.8 against a baseline of 6.7. A new batter does not change the gear. It applies the brake.

CONTRARIAN: EXCELLENT DATA CAN STILL TELL THE WRONG STORY

The tables invite an easy conclusion: more dots here, fewer runs there, therefore the batting is poor. But correlation is not causation, and I have been burned by that error before.

Three alternative explanations survive scrutiny. Environment first: slow, spin-friendly Mirpur pitches raise dot-ball rates in the middle phase for every side. Seven of my ten home matches are in that condition while the baseline sides play a larger share abroad. Home surfaces add three to four percentage points of dots to any baseline — normal. Part of the 11.2-point gap is environmental. Not all of it.

Strategy second. With two or fewer wickets down, Bangladesh score at 119.2 in the middle phase; after three wickets fall, 96.8. That lower figure may not be foolishness. If six of the eleven cannot sustain 140, then 96.8 might be the highest available winning probability on a given night. I am not claiming that. I am claiming the table alone does not settle it.

Selection and role third. Six batters cannot all be given the same job. In my compilation, the player cast as the middle-phase anchor defends 31 percent of deliveries in that phase. The role was never defined; only the team result was used to point a finger.

And a trap in my own baseline-first habit: it can flatten genuine exceptions. Jaker Ali's 50 from 34 balls in a middle phase may sit outside the baseline as an outlier — and the lazy move is to discard it. Some single-match exceptions are real. So I publish z-scores alongside the mean, so the exception gets its own place. An innings of 32 from 9 balls in the middle phase against India is not an average; it is a match-state decision. A baseline explains it. It cannot reproduce it.

One more warning, aimed at myself: the ten-match threshold is a human construction. Where a batter faces more middle-phase balls on spin-dominant surfaces, fixing the threshold at ten is defensible-looking and analytically weak. The threshold is a rule of intent, not a claim of power. Written down and then ignored, it stops being a rule.

TAKEAWAY: WHAT I WILL WATCH NEXT CYCLE

Three specific observations, nothing more.

First, the first twelve balls after the tenth over, regardless of the left-right combination. If the run rate stays under seven an over there, the problem is tempo, not technique.

Second, sweep coverage in the spin match-ups of overs 7 to 11. If fours and sixes against off-spin do not rise, the death overs will keep producing fine-leg hoists instead of point-side placement — where there is no boundary, no six, only a fielder.

Third, strike rate in the five overs after a wicket falls. If that number stays under 100, the next cycle will hide this flaw behind a pendulum: slow middle overs, spectacular death overs. A few series will be won on exactly that pattern. Then a semifinal will end at 94 for 5 off 43, and the ledger will have said it years earlier, to nobody listening.

Wednesday's innings will end with a six or a stumping. Either way, I will keep the ledger open on the six overs where the light goes out — because for ten years that is where Bangladesh's T20 story has actually been written.