The Quiet Economy of Dot Balls: The Powerplay xG Nobody Watches in the BPL

Last week at Mirpur I was watching a powerplay. The scoreboard said 54/1 in five overs — six fours, one six, a wave of applause around the ground. My notebook was keeping a different story: behind those 54 runs were fourteen edges, three mistimed pulls, and only two shots I could call clean contact. In the same overs the opposition's new-ball seamer was bowling four to five inches outside off, with almost nothing but slower balls. The scoreboard and the shot quality were telling two different truths that evening, and nobody was writing the second one down.

Watching matches for years has taught me one thing: in cricket, runs are an outcome, not an explanation. A side that takes 55 in the powerplay may not have batted well at all — what tells you that is shot quality, the line and length it faced, and its intent against the field. This is where a league can see its own reflection — if someone holds up a mirror. And holding up a mirror is not just collecting data; it is asking data the right question.

Context — the league, the pitch, and the data vacuum

The Bangladesh Premier League began in 2026, and it has carried an odd duality ever since: the most passionate audience in the subcontinent, and the least structured data. The ground has a scoreboard but no stroke map; it has a strike rate but no account of how risky each shot was. It was to fill that vacuum that in 2026, at twenty-four, from my room in Rajshahi, I joined Golpo Sports and set out to build a league-level expected-value model. I coded 1,248 shots, logging line, length, shot type and contact quality for each.

The result was striking. I was still writing the word 'deserved' back then. After that the writing changed — 'deserved' out, 'xG differential' in. Shot quality entered every match report, not just possession. In Bangladesh I taught a league to see its own xG — the way I had taught it in football, the same way in cricket.

Building cricket's xG is harder than football's, because here a shot and a run are not the same. In football a shot either goes in or it doesn't; in cricket a shot becomes one run, four, six, or a catch. So I began with a simple question: in the powerplay, what shot, against what ball, deserved how many runs?

By the way, the first proof of how much shot quality matters came on 11 September 2026 in Johannesburg — Chris Gayle's 117 against South Africa was the first T20 international century. Many of those shots flew beyond the boundary, but the context was a helpful wicket and thin fielding. From the very start, cricket taught me: runs and shot quality are two separate accounts.

Core — a three-pillar model

The model stood on three pillars.

Pillar one: powerplay expected runs (xR). I classified every ball — line (off/middle/leg), length (short/good/full), speed — and the batter's shot type. Then I computed how many runs that ball was fairly worth on average. It showed that powerplay totals often inflate on luck: an edge for four, a misfield, or a top-edge six are worth little in xR, yet look identical on the scoreboard. When a batter like Shakib Al Hasan plays over mid-off, that is skill; when the same runs come through the edge, that is mere fortune — the model separates the two.

Pillar two: the dot-ball pressure index. Here my football life paid off. At the 2026 Russia World Cup, in Germany vs Mexico, Germany's PPDA was 6.9, which showed they were pressing but weak in transition. PPDA showed me Germany — and showed me that pressing is a structural language, not the name of a game. In cricket I translated that language: batting press. In the powerplay, the share of balls a side plays inertly (dot or single) is its dot-ball pressure. A side that strings together two or three overs below 6 in pressure is not really building a score — it is cracking.

Pillar three: adjusted death-over economy. In the last five overs, runs conceded are not really the bowler's skill but the bowler's risk management. Mustafizur Rahman's cutter, when it becomes a yorker, is a different weapon; when the slower ball drifts into the batter's arc, the scoreboard swells. So I measured death overs by successful deliveries, not by economy.

Across all three pillars I looked back at a league season. Where one side scored more, xR said it had actually got more than expected; where it scored less, it was underperforming. That gap is what a coach calls form and what data calls regression. An ESTJ builds the pipeline first and the poetry second — so I set up the collection process first, then wrote the story.

I remember one particular match. In the first innings a side made 48/0 in the powerplay — it looked superb. But ball by ball, 18 of those 48 came through edges threaded between two fielders, and eight came off free hits from no-balls. xR said those overs were fairly worth only 34. In the next match the same batting line-up made 52, but this time xR was 50 — genuine skill. The two scores were nearly identical; the stories were completely different.

Contrarian — a mirror, not a god

This is where caution belongs. A metric is not truth; a metric is a question-making device. I write a hypothesis down in advance — if a side's powerplay xR is higher, will it win more matches? The answer is often: no. Because correlation is not causation. Bangladeshi pitches are scratched and bare, spin slows through the day, and dew in the second innings overturns every calculation. A model that ignores pitch and dew is imported dogma, not local reality.

Another trap: assuming data infrastructure exists. In the BPL there is no reliable stroke map for every ball; data has to be built by hand — by scorers, coaches and video analysts. So a model can never be a substitute for a decision, only a mirror for it. Empty stadiums taught me that home advantage is a variable, not a law — and in the same way the BPL's home favour shifts every season, and must be measured, not assumed.

Dressing-room constraints matter too. On a slow pitch a captain may reduce his powerplay aggression because his seamers' yorkers do not grip there. That decision is not a failure of data but a limit of data — which is why the model has to be calibrated with the coach in the room. I do not want a model telling someone 'this batter is bad'; I want it telling them 'this shot is this risky in these conditions.'

Takeaway

For the next round, one signal for the selectors: a scoreboard 50 and an xR 50 are not the same. Those with high dot-ball pressure in the powerplay have soft runs; those with low pressure have hard runs. Knowing the difference makes selection less emotional and more reasoned. The question now sits with the selectors, not me: will you look at a batter's form, or at his shot quality?