The Middle-Over Ledger: 41.3 Percent Dots in 11,112 Deliveries and One Wrong Pairing

Over the last four matches, the league's run rate between overs seven and fifteen has fallen from 7.84 to 6.41. The powerplay is effectively flat—8.12 to 8.06. The death overs are flat too—9.74 to 9.81. Right in the middle of the regular season, one window has narrowed, and that window is the middle eight overs.

Last Tuesday, in the fourteenth over, a left-arm off-spinner bowled four balls and three were dots. In the next over, the same bowler, the same line, two runs. In one match that is an accident. Across five matches it is a pattern. I had sat down to count fast bowlers' workloads; at this stage of a season that is the obvious job. When I opened the ledger I found something else had changed—and it had changed not in the speed of the ball but in where the fielders stand.

This piece is the accounting of that pattern.

The Grammar of the Ledger

Match and shot counts | Context: opening a long-form data retrospective.

The arithmetic is simple. Forty-six matches of the current regular season, 11,112 legal deliveries, up to Wednesday night. Of those, 44 matches have ball-tracking coordinates in my archive; two were coded from broadcast alone, with no pitch map. So the turn and deviation figures below come from 40 matches—six matches could not be reconstructed because of camera cuts or the absence of an overhead angle.

I log every ball in five fields: length, line, fielding positions at the moment of delivery, the batter's shot type, and the outcome—dot, single, boundary, wicket, extra. Without those five fields no conclusion is reachable, because there is no such thing as an over that went well. What went well was which ball, against whom, into which field.

The Khulna ledger did not lie: 132 matches, 2,847 shots, and one quiet conclusion. From that 2026-18 ledger I borrowed one habit—writing down the number I do not want to see. So let me state it here: before the season began I predicted powerplay aggression would drop, on the reasoning that with two fielders out during the new-ball phase, bowlers would use the cutter more. The prediction was wrong. The powerplay run rate went from 8.12 to 8.06—in statistical terms, zero. The null result stayed in the ledger, because without it the rest of the accounting would be nothing but a story.

I keep the length bins deliberately coarse. There is a reason six-to-eight is separated out: in T20 that zone is the lowest-risk delivery, and it is where batters make the most mistakes.

I code shot type in four bins—defence, rotation, lofted, flat. Premeditated shots are flagged separately by footwork; if the back foot has already gone back before release, I treat the shot as decided in advance. That flagging carries an error rate, probably 10 to 12 percent. I have to carry that error alongside every number I quote.

What Is Happening in the Middle Overs

Between overs seven and fifteen, the dot-ball rate has risen from 34.8 percent to 41.3 percent. The boundary rate has fallen from 13.2 to 10.1. Singles per ball have gone from 0.42 to 0.36. Spin's share of deliveries in this eight-over block has risen from 62 to 71 percent.

The real key to the middle-over scoring drop is not fewer boundaries but more dots. The difference is one of accounting. If a side loses one boundary an over, that is four runs. But when two extra dot balls appear per over, they do not merely remove runs—they consume balls, shrinking what is left in the batter's hands before he reaches the death overs. A 41 percent dot rate means roughly two and a half balls an over pass with no run at all.

That raises a question the league table cannot answer: are these dots the bowler's skill or the batter's decision? I split it ball by ball. On six-to-eight length, on the stumps at top of middle, the dot rate has risen most—from 31 to 39 percent. But the attempt rate on that same length has also risen, from 54 to 61 percent. The batters have not stopped. They have chosen the wrong shot.

Which shot? The sweep, and the reverse. In the middle overs the success rate of premeditated sweeps has fallen from 58 to 41 percent. On bouncer or fuller length the rate is essentially unchanged. The damage is specific: the decision to sweep against spin.

There is one more thing television does not show. In the middle overs, sides are not hunting a boundary every two balls; they are hunting one boundary an over, and they wait until the sixth ball to try. So the first three balls of an over carry low risk and low return. The cost of that waiting lands at the end of the over, when the finisher walks in with only 11.8 balls left.

The Geography of the Dot Ball

Data Monk archetype and the cricket domain | Context: beginning a data-driven match analysis.

More important than where the ball goes is who is standing where. At the moment of delivery, the share of balls with six fielders inside the thirty-yard circle has gone from 38 to 51 percent. More ring fielders change the arithmetic of the gap-finding stroke—that is nothing new. What is new is which fielder.

Teams have kept cover and midwicket, and pulled point back to long-off and deep cover. So the straight drive and the late cut—the two lowest-risk strokes—are being converted into singles, while no space is left in the deep for the pull or the slog sweep.

Here is a numerical proof. Sides that pulled deep cover back and pushed short third up conceded 7.1 an over in the middle overs; sides that did the opposite conceded 6.2. That is 0.9 runs an over—seven runs across eight overs. In T20, seven runs changes results.

There is another layer of field placement the scorebook never shows: the change within a bowler's own over. The same spinner bowls his first three balls to a cover-heavy field and his last three to a deep-heavy field. The bowler himself knows where he will be hit, and has priced it in advance. If the batter cannot read that pattern, his first three balls are dots and his last three are singles—an over that ends with six runs.

The Left-Hander Fracture

Sides that promoted a left-handed batter at number four in the middle overs scored at 7.62 in that phase; sides that did not scored at 6.18.

I did not believe this at first, because the sample is small. Across 46 matches, the left-hand-promoting sides appear in only 11. Even so, a mechanism holds. Seventy-one percent of middle-over deliveries in this league come from off-spin and leg-spin. For a right-arm off-spinner, a left-handed batter means a different arc, a different line, a different field. The ball is not turning—but where it lands is changing. Force a spinner off his favourite angle and runs come in the first two balls of the over.

The inverse is clearer still. When two anchors are together at the crease—both batting below a 110 strike rate—the middle-over run rate is 5.94. That pairing has formed 23 times this season, and every time the dot rate over the following four overs has gone above 45 percent. Two hitters together produce 8.4, but those pairings last 2.1 overs on average before a wicket falls, and the side loses the advantage of scoring 9.8 in overs 16 to 20.

There is a reason I look at the twelfth over separately. Within the seven-to-fifteen block, that over is the moment both ends of an innings are visible at once—the set batter and the new one. The run rate in that over is 6.1, while in over thirteen it is 6.9.

The Number That Did Not Move

Turn. By the pitch maps, deviation off the stump line has fallen from 4.1 degrees to 3.9. Spinners' revolutions per minute are essentially flat. The ball is not turning more. The pitch has slowed down—that story finds no support in my ledger. What changed was the field, the shot selection, and the role definition in the batting order.

I checked the second innings too, wondering whether dew was at work. In the middle overs, the first innings run rate is 6.3 and the second innings 6.5. The gap is small enough that dew cannot be called a cause. That is another null result, and it stayed in the ledger.

I checked the toss and the order of innings as well. Sides batting first scored at 6.4 in the middle overs; sides batting second also 6.4. No difference at all.

This is where the most important caution sits. The middle-over scoring drop and the mid-season transfer window happened in the same week. Three teams replaced their number four; two brought in overseas spinners. Correlation is not causation. The three teams that changed players saw their middle-over run rate fall from 7.3 to 6.5; the teams that did not fell from 6.8 to 6.3. The gap is small. The transfer window is not the cause here; it is merely contemporaneous.

The Anchor Is Innocent; The Pairing Is Guilty

Transfer administrator's ledger | Context: accounting for mid-season squad changes.

In this league, the number three position receives 24.6 balls on average, at a strike rate of 108.4. Read that number alone and the easy conclusion is: the anchor is slow, move him. My accounting says that conclusion would be wrong.

The reason lies in the balls. Between overs seven and fifteen, the league's average strike rate is 128. If those 24.6 balls went to an aggressive batter, roughly 18 extra runs would arrive. But the match accounting does not stop there. Look at the ball-consumption pattern of a batter striking below 110—he takes the first four balls of an over, and the scoring still does not accelerate toward the finisher. So the side's best hitter arrives at 16.3 overs and gets only 11.8 balls.

The problem is not the role; it is the pairing. One anchor and one set hitter produce 7.8 in the middle overs. Two anchors produce 5.94. Two hitters produce 8.4, but wickets fall quickly, and that side forfeits the 9.8 an over available from overs 16 to 20.

This is the counter-intuitive part, and it is my least comfortable conclusion: the league's middle-over problem is not the fault of any single player. If the coach's sheet has a box marked anchor and a box marked finisher, and in reality nobody occupies the space between the two boxes, then overs seven to fifteen have to be dragged by one man alone. No one can.

On bowler-batter matchups I do not write names, I write roles. Because in 46 matches, the sample for a single bowler-batter duel might be eight balls. Judging a person on eight balls is not the work of a ledger; it is the work of a smear.

Fixtures and Load: The Middle-Over Spin Block

There is another layer here, which I found while counting workloads. Spin's share of middle-over deliveries rising from 62 to 71 percent is both a tactic and load management. Sides playing three matches in five days have given spinners more of the middle overs than seamers—73 percent on average, against 66 percent for the rest. Seamers' overs are being rationed.

That is rational, because in this schedule seamers carry the highest soft-tissue risk. However good the medical team, nobody absorbs a weekly load of three matches in five days—that much is clear from my numbers. But there is a side effect. As the spin block grows, the middle overs slow further, because spinners are bowling with the older ball, not the new one, and the field stays set. The scoreboard pays for the decision to protect the load.

The Error Log

A ledger is only complete if the errors are written into it, because a model without an audit is just an opinion.

First error: I predicted that the side which signed two overseas spinners after the transfer window would see its middle-over run rate rise, on the reasoning that batters would have less information against unfamiliar bowlers. In fact its run rate fell from 6.9 to 6.2. The mechanism worked in reverse—teams have not yet read the new spinners, and the batters stayed extra cautious.

Second error: I assumed the advantage of promoting a left-hander at four would persist into the last five overs. It did not. In the final five overs the left-right difference is 0.3 runs an over—inside the statistical noise.

Third error was methodological: in the first two weeks I split ball length into only three bins. The dot-ball pattern only appeared once I separated six-to-eight length out. Change the bins and the answer changes—the most uncomfortable truth about working with data.

What this ledger cannot say is also clear. I do not know a player's sleep the night before, his injury, or his personal circumstances. Forty-six matches cannot make any conclusion permanent. I have used this method in football before, and the transfer to cricket is not direct. Football's xG measures shot quality; cricket's equivalent measure is ball length and field geometry. Different names, the same question: how good was the opportunity, and how bad was the decision.

A Key for the Next Three Matches

The Khulna ledger did not lie, and this ledger is not saying the league's batting is poor. It is saying the decision point has shifted one step—from the batter to the coach's sheet.

Over the next three matches I will count two things. One, who bowls the twelfth over—in the middle overs this season, spin there has cost 8.4 an over, seam 6.9. Two, the number four's first ten balls—for a left-hander the strike rate is 136, for a right-hander 112.

These two numbers will say more about who rises over the next three weeks than the league table currently does. And if a spinner comes on in the twelfth over again, and a right-hander walks out at four again, there is no need to open the ledger—the answer is already written.