The Uncounted Innings: Why Dot Balls in Overs 7–15 Are South Asia's Cheapest Number

Hook

Late last season at Mirpur I was reading 186 on the scoreboard and writing 59 into my notebook. Fifty-nine dot balls in 120 deliveries. Nearly half the innings produced no run at all, and the side still made 186. The three-minute highlight package contains no frame of any of those fifty-nine. It is built from six sixes and two diving catches.

After the match I turned the page and found thirty-four of the fifty-nine had fallen between overs 7 and 15 — the phase broadcast calls "consolidation." The scorecard logged dot, dot, dot. The narrative logged "the batting side was under pressure." Both are true. Only one of them got a price attached.

After three seasons of watching from the stands and rewatching on screen, the conclusion I keep arriving at is simple: if a T20 scorecard is a lossy compression of the match, the dot ball in overs 7–15 is the largest discarded file. And across the 2026–2026 franchise auctions, the auction price shows effectively no relationship to it.

Context: method before story

Let the ledger breathe before the narrative does. Definitions first.

Dot-ball rate (DB%) — share of deliveries in a given phase that produced no run. Rotation rate — the ratio of one- and two-run balls per over, a proxy for strike-changing ability. Role-adjusted strike rate (RASR) — strike rate re-based against the scoring environment of each phase, because overs 16–20 and overs 7–15 are not the same run-scoring world.

The sample is my own ball-by-ball coding: 202 innings across BPL 2026, 2026, 2026 and IPL 2026, 2026, 2026 — 24,240 deliveries, hand-coded from the television feed. The sheet is open. Anyone may argue with it, provided they code the same definitions first.

The market context matters too. The IPL auction sits in Kolkata, the BPL auction in Dhaka, the PSL auction in Lahore. The same batter is bought at three different prices, and each room speaks a different valuation language: Kolkata buys "match-winner," Dhaka buys "experience," Lahore buys "power-hitter." All three are reading the same file, just different pages of it.

One more thing. When I tracked 92 Bundesliga matches behind closed doors in 2026, home win rate fell from 43.3% to 33.3%. The stadium was empty; the numbers were not. Cricket teaches the same lesson: until you hold pitch, ball and time of day constant, you cannot separate dot-ball noise from dot-ball skill.

Core: what the scorecard discards

Layer one, the aggregate. Middle-overs (7–15) dot-ball rate in my sample: BPL 41.6%, IPL 34.1%. A 7.5-point gap that invites the lazy conclusion that BPL batting is simply worse. Before jumping there, go one layer down.

Layer two — a dot ball is not one thing. I fixed three categories before looking at any outcome, because the more bespoke roles you invent, the more every cheap batter looks like a discovery only you can see — and that is your artifact, not the market's.

D1, ball-won dot: the batter was beaten, or defended a genuinely attacking delivery. The bowler's asset.

D2, ball-lost dot: slot ball, full toss, short and wide — and the batter failed to convert it. The batter's cost.

D3, structural dot: no rotation option existed — the set batter was at the non-striker's end, or a lower-order batter was in the middle of the innings.

My sample split: BPL D1 46%, D2 27%, D3 27%. IPL: D1 52%, D2 19%, D3 29%. Read that carefully. The skill gap between the two leagues does not sit in D1. It sits in D2. Bangladesh's batters are not being beaten more often; they are failing to convert easy balls into runs. That is a technique problem, not a temperament problem — and technique can be measured, whereas temperament cannot.

Layer three — the non-striker's overs. This is where my interest actually lives. I measured "stranded overs": an over in which one batter faced five or more deliveries while a set partner with a strike rate above 130 stood at the other end. In the BPL 2026–26 sample, 18.4% of middle overs qualified. Those overs averaged 6.2 runs. All other middle overs averaged 8.9.

The cost is 2.7 runs per stranded over. Two stranded overs in a typical match means five or six runs leaking out silently, never entering any conversation because they carry no news value. I count the silence between the deliveries.

A finisher in the Jaker Ali or Rinku Singh mould is valued for the death phase, where the per-ball run baseline clears nine. The team's real leak happens earlier, when the set batter is not on strike. Nobody measures that leak, because the camera stays on the set batter's face.

Layer four — role-adjusted strike rate. In my sample, overs 16–20 score at 10.8 an over; overs 7–15 score at 7.6. A No. 5 whose balls fall 60% in the death phase carries that phase's subsidy inside his raw strike rate. RASR strips it out. For a No. 3 or 4 in the Towhid Hridoy or Suryakumar Yadav mould, the arithmetic runs the other way: they face more balls in the low-scoring phase, so raw strike rate undersells them. A franchise buying purely on raw strike rate will systematically overpay for the first profile and underpay for the second.

Layer five — auction price. I matched 74 IPL and BPL buys between 2026 and 2026 against those metrics. Price vs raw strike rate: r = 0.58. Price vs RASR: r = 0.61. Price vs middle-overs DB%: r = −0.14.

That last number is essentially zero. The market pays for overs 16–20 and does not price overs 7–15 at all. Consider two batters: one with a raw strike rate of 135 but a 44% middle-overs dot-ball rate; another at 128 with a 34% dot rate. In my sample the first commanded the higher median price. On modelled run value, the second is ahead.

This error is invisible in the Heinrich Klaasen tier, where boundary rate swamps the dot-ball cost. It shows up in the middle band — squad numbers six through ten — where mispricing compounds into crores over a five-year cycle.

Layer six — mispricing between markets. Here the story crosses a border. The same batter fetches different money in Dhaka and Kolkata, justified by raw DB%. The raw BPL–IPL gap in my sample is 7.5 points. After conditioning on pitch type, time of day and ball age, it narrows to 2.1 points.

So six of those seven points are pitch and ball-condition noise, not technique. A franchise that discounts a Bangladeshi middle-order batter on raw numbers is discounting geography. That is arbitrage, not evidence of weakness. Morocco conceding 0.89 xG per 90 in the knockout rounds taught me that defensive quality can only be read once the environment is held fixed. In cricket the environment moves faster.

Contrarian: maybe my own number is shopping for a story

Now against myself.

First, correlation is not causation. A high middle-overs dot rate does not prove batters are playing badly. Slow pitches, reverse swing on a used ball, evening dew under the lights — conditions generate dot balls as easily as batters do. My taxonomy tries to separate them, specifically by splitting D1 from D2, but a constructed boundary can look cleaner than the real one.

Second, here is the base rate for my own overrides: between 2026 and 2026 I publicly challenged 31 consensus positions. Seventeen proved right, 14 wrong. That is 54.8% — a coin toss. My skepticism has no demonstrated edge. Anyone leaning hard on my argument should know that.

Third, the sample is a convenience sample: televised matches only, one camera angle, hand-coded without ball-tracking. Confidence intervals on small sub-slices will be wide, and I am writing that down rather than hiding it.

One claim survives anyway, because it is narrow: a dot ball is not evidence of a batter's quality, but the way it relates to runs is not the way the auction relates to it. Wrong price and wrong valuation are separate things. I am only claiming the second.

Takeaway: what I will watch over the next twenty matches

A narrative can be written in one evening. A falsifiable record takes longer. So this is pre-registered, with explicit thresholds, and I will grade it in public — win or lose.

Pre-registered prediction: across the next twenty BPL matches, sides holding a middle-overs DB% below 38 will score 160 or more in over 65% of innings. Sides above 44% will do so in under 40% of innings. Grading date: seven days after the final match of the next BPL cycle.

I am filing this on time rather than perfectly. An imperfect record published on deadline beats a flawless record published after the news cycle has moved.

The scorecard will tell you who won. The highlight reel will tell you who dazzled. Neither will tell you what quietly vanished between overs 7 and 15, or what it was worth. So the next time someone celebrates a 45 off 35, I will ask one question: in how many of those 35 deliveries did the man at the other end not touch the ball at all?