Overs 7 to 15: The Spell That Loses the Scorecard and Wins the Match

Last round I put two lines from the same scorecard side by side. Bowler A: four overs, 22 runs, no wicket. Bowler B: four overs, 38 runs, two wickets. By the end of the night, B was the story. I am not diminishing him. I am saying the two columns are one event written in two languages. The wickets column sits in front of the camera; the economy column sits behind it. So the spell that pinned a chasing side through the 14th over, the one that made the collapse in the next two overs possible, never gets a name. For more than three years I have run a small model on phase-adjusted economy and a pressure index. I built the Burnley model to hear the mean, not to cheer for it. Cricket gets the same discipline. The premise is simple: the three phases of a T20 innings are never priced equally. A wicket in the powerplay brings the best batters to the crease, so the damage is contained. A wicket at the death folds the entire innings. The middle overs, seven to 15, are effectively a tax office: run value per ball is at its lowest, dot-ball value at its highest. Nobody reads that ledger, because the ledger has no wickets in it. My model takes four inputs. One, a venue-adjusted run baseline, meaning what this ground usually produces in this window of the innings. Two, phase-specific economy. Three, dot-ball percentage. Four, how far the opposition's run rate falls in the two overs after a given delivery. That last input does the real work; the other three are bookkeeping. In a regular season this model earns its keep, because league-table pressure has not yet become a headline. Two confessions up front. First, most of my data is English county, Edgbaston and Trent Bridge market prices, so my lens is partly European; dropping the same coefficients onto Asian conditions produces errors, and that error happens daily in my industry. Second, this is not an advertisement for a single innings. Where the sample is thin, I print a range. A model is a confession of what you refuse to guess. I used to open articles with the scoreline; now I open with the gap between the model and the market. Across five years of T20 data, one pattern holds: where a side's dot-ball percentage in overs seven to 15 crosses 40, its final total lands roughly 10 to 14 runs lower. Those 10 to 14 runs decide matches, and they never appear in the post-match discussion. Jasprit Bumrah's IPL economy sits near seven, and everyone knows that. What fewer know is the variety he carries from the 16th over onward: four distinct slower balls from one release point. The market prices that variety because it is visible. What is not visible is the finger spinner and his dot ball. In Asian conditions, a finger spinner's dot-ball percentage between overs seven and 15 frequently sits in the 35 to 45 band in my numbers. The reason is geometry, not magic: on a slow, low, turning surface the batter must take the risk, and the spinner need not. A fourth-day Mirpur surface, Colombo humidity, the Galle cover: three venues, one rule. Mehidy Hasan Miraz's Test bowling average at home is roughly half of his average abroad. That gap is a story about conditions, not about skill, and it tells you his away average is a noisy number. Shakib Al Hasan's 700-plus international wickets are ten years of accumulated capital, not a single-night arrival. The Croatia position was never faith; it was a mispriced midfield. In cricket the finger spinner is that midfield: the position where the fee is lowest and the job description is heaviest. Mustafizur Rahman is priced on his cutter and on death-overs memory; 17 wickets and an Emerging Player award in the 2026 IPL built that memory. In my three-year data, his most valuable ball sits between overs nine and 14, where batters cannot line up the slower one and the scoreboard quietly stops moving. We buy memory and we measure the present, and that gap is where I work. Now the part where my own model is weak. A spinner's economy in overs seven to 15 measures more than his skill; it is also a function of the opposition's risk calculus. When the batting side is losing wickets, his numbers look beautiful; when a set batter is in, the batter chooses to attack. Separating intent from execution is not data work, it is eye work, and I do not skip the eye work. Correlation and causation are different objects, and the analyst who forgets it has a model that breaks in three weeks. The second trap is home advantage. After stadiums emptied in 2026, home advantage in European football fell by roughly ten percentage points; when the crowd left, the edge left with it. Asian cricket never showed that collapse, because home advantage there does not sit only in the crowd's head. It sits in pitch preparation, in spin coaching, in the familiar rhythm of domestic bowlers. Empty the stands and the edge survives, because it is a condition, not an atmosphere. Which means numbers built outside Asia and numbers built inside Asia cannot be weighed on the same scale. That is exactly why I rewrote the model in 2026. One more thing the model will not hold but should: bowling load. Shakib, Mustafizur, Bumrah — these bowlers rotate between franchise and national duty all year, and the cost of injury is financial as well as physical. A bowler whose phase value is highest needs the most careful scheduling. A model that treats a bowler as a six-multiplier vector commits its offence off the field. Next round I will watch two numbers closely. One, venue-adjusted dot-ball percentage in overs seven to 15. Two, the distance between that number and the award committee's pick. Where the distance is widest, a squad-building decision is wrong, usually at the auction table. The market reacts to stories; I wait for the residuals to speak, because that is when buying becomes possible. The question now is this: at the next auction, who buys a finger spinner who takes two wickets in ten matches, and who settles for the commentary booth's version of him?