Mirpur's Dot-Ball Audit: Overs 7–15, Not Powerplay Runs, Write the Result

Last month I reopened the scorecard of a T20 match at Mirpur's Sher-e-Bangla National Cricket Stadium about an hour after the game ended. What the eye misses in real time tends to surface in a ball-by-ball log. The home side had 51 for no loss at the end of the powerplay. The commentary called it a commanding start. In the same innings, their dot-ball rate between overs 7 and 15 was 46.8 percent. They needed 62 off the last five and finished 49 short. Thirteen runs.

That is the anomaly. In Bangladesh's home conditions, the link between powerplay runs and winning is far weaker than the broadcast narrative suggests. In my log, teams passing 50 in the powerplay win 52 percent of the time. Teams that score 42 but keep their 7–15 dot-ball rate under 30 percent win 68 percent. The very column we celebrate every over-break is the weakest predictor at this venue.

Context: start with the pipeline, not the prediction

When I built a standard logging template for the Bangladesh Premier League in 2026, I trained three Khulna-based interns to enter every ball — shot line, length bucket, shot zone, runs, dot flag. Match prep dropped from nine hours to two and a half, because I stopped assembling matches from memory. That habit moved straight into cricket.

Every entry in my current structure carries: match ID, venue code, innings, over, ball number, batter, bowler, pitch age, dew flag, toss result and a phase tag. The phase split stays fixed — powerplay 1–6, middle 7–15, death 16–20. When an impact sub arrives or DLS shortens an innings, I re-tag the phases rather than writing over the old ones.

Definitions stay fixed too. Dot-ball percentage is dots divided by legal balls faced in that phase. Rotation rate is singles plus twos per legal ball. Boundary conversion is fours per ball in a given zone. Phase-adjusted run rate is runs above or below the venue-and-phase par. If a definition shifts mid-tournament, the model collapses — which is why a clean match ID is worth more than a clever model.

Cleaning cricket data is harder than cleaning football data. Football has few events across 90 minutes; cricket has six separate entries per over, plus DLS revisions, impact subs, credit allocation on run-outs, and separating wides from no-balls. My rule is blunt: if the scorecard and the ball-by-ball feed do not reconcile, the match leaves the log. In rain-shortened series I have watched phase comparisons become meaningless simply because an eleven-over innings was still tagged as a full one.

Venue separation matters. Mirpur, Chattogram's Zahur Ahmed Chowdhury Stadium, Sylhet International and Khulna's Sheikh Abu Naser Stadium behave differently — bounce, outfield speed, square boundaries. At Mirpur the ball grips once it softens, the outfield is slow. Comparing raw boundary counts with Wankhede or Chinnaswamy is not a comparison at all. The real India–Bangladesh lesson is this: the same metric carries two different meanings in two places, and analysis that does not change with the venue is not analysis.

Core: the data evidence chain

My log holds 64 T20 matches at Mirpur between 2026 and 2026 with clean innings-level phase data. I have re-run four relationships repeatedly, and they keep landing the same way.

First, powerplay run rate against win percentage — a coefficient of 0.21. Fast scoring in the first six overs does push a team toward a win, but the relationship is so weak it cannot carry a claim on its own. Television match summaries devote the most space to exactly this column.

Second, dot-ball percentage in overs 7–15 against win percentage — minus 0.58. That is the strongest relationship in my log, roughly three times the strength of the powerplay number, and it comes with the tightest confidence band.

Third, death-over economy against win percentage — minus 0.44. Expected, but subordinate to the middle phase. Fourth, the toss. Chasing sides at Mirpur win 54 percent overall, but that climbs to 61 percent when the dew flag is on.

The mechanism is mechanical. After roughly the 12th over the ball softens, spin gains weight, and spinners start conceding under two per legal delivery. A side that rotates strike between overs 7 and 15 — one run a ball, moving fielders every second delivery — needs fewer runs at the death, faces less strike-rate pressure and preserves wickets. A Mirpur match is not won or lost in the powerplay; it is written in the bookkeeping of overs 7 to 15.

At player level my log tells the same story. Litton Das plays the fielders rather than the boundary in that phase, which is why his dot-ball rate under pressure stays low. Towhid Hridoy works the ball into the point and third-man gaps, and that is precisely why his scores rarely look spectacular while his team wins. Gifted batters who arrive with a pre-planned pull shot eat dots the moment a spinner changes length — and the match decisively turns there.

On the bowling side, Mehidy Hasan Miraz and Rishad Hossain explain this innings-level relationship better than anyone, because they build pressure with dots rather than wickets. Mustafizur Rahman and Taskin Ahmed post good death-over economy, but if the foundation from overs 7–15 is missing, the game has already left before the ball reaches their hands. This is the seam where both pre-tournament forecasts and betting valuations go wrong.

So, is this strong enough to change coaching instructions? My answer is a conditional yes — but I have to write the condition myself.

Contrarian: correlation is not causation

Here I have to argue against my own instinct. A minus 0.58 link between middle-phase dots and winning does not prove dots cause defeats. The 2026 empty-stadium period was a control group we never requested, and it taught something uncomfortable: environmental variables routinely masquerade as skill.

Three alternative explanations sit open in front of me. One, teams facing weak spin attacks naturally record fewer dots — the metric is a matchup proxy. Two, when two matches are played on the same day the second pitch is older and dew arrives, which lowers dots — the metric is a pitch-age proxy. Three, in DLS-reduced matches the phase boundaries move, so the comparison is contaminated before it starts.

None of these cancels the finding, but each shaves my confidence. So I write down what evidence would change my mind: if venue-age stratification pulls the coefficient from 0.58 down below 0.25 and the confidence interval touches zero, I demote middle-over dot ball from primary to supporting indicator. If it survives, I keep revising the framework. Every outlier is a question the data is asking you — and dodging the question is the real delay.

Betting markets need the same discipline. Pre-match talk circles powerplay run lines and six-over strike rates, because that is where the story and the crowd are. The edge hides in the boring columns — the 7–15 dot ratio, the singles outside the fielding restriction, the post-toss dew flag. Information everyone can see is priced in long before you get there.

DLS consumes even more time than cleaning. When rain arrives, moving a phase boundary is not a scorecard correction; it rewrites the basis of comparison. Without an audit trail, no explanation of those moments holds — skip the bookkeeping for chaos and you will mistake chaos for skill.

Takeaway

Across the next four or five home series I will track three things together: the bookmaker's player quota, the dew flag, and the home side's spin pair's middle-over dot rate. When all three align, I will update my pre-match model on 7–15 rotation rate, not on powerplay score. That may be less comfortable for a coach, because there is nothing glamorous to show in those overs.

The question worth putting to a coach or a selector is this: which part of your innings is actually the most expensive? Not the segment the scoreboard advertises — the one the log quietly accumulates. And if you are not keeping that account, your story may be compelling, but it will not be auditable.