The Open Ledger: T20's Uncounted Overs and the Two Prices of One Bowler

One night last April, after an IPL match had ended, I closed the scorecard and opened a separate ledger. A left-arm cutter's figures: four overs, 22 runs, no wickets. The highlights package does not contain a single ball he bowled. His name is not spoken once on commentary. Seven weeks later, in the auction room, nobody raises a paddle for him.

My ledger said otherwise. Of those twenty-four balls he bowled, fourteen were dots. Eleven of them came inside the powerplay; three came in a half-dead over, when the opposition had six wickets in hand and needed eleven an over. The scorecard records those fourteen balls simply as 'the rest'. Let the ledger breathe before the narrative does.

A cricket scorecard is a lossy compression. Every ball contains at least thirty decisions — the bowler's line, the batter's intent, a fielder shifting two yards, an appeal dissolving. Six columns retain a fraction of it. My job is to rebuild what was discarded. So the question is not 'who is a good bowler'; the question is which overs nobody counts, and why nobody counts them.

Method first, verdict later. I used 1,840 overs I logged by hand across three T20 leagues between 2026 and 2026 — ball by ball, cross-checked against commentary and the scorecard. The window was fixed in advance, long before outcomes were visible. The sample is limited and my limitations are explicit: ball-by-ball public data is not equal across leagues, there is no clean way to control for fielding quality, and small samples produce wide confidence intervals. What I can do is hold phase baselines constant.

In my log, the powerplay (overs 1-6) league average is 7.9 runs per over. The middle (overs 7-15) averages 8.1. The last four overs average 10.2. Those three numbers sit on the same pitch, in the same month, against comparable bowling attacks.

Now consider what an economy rate actually is. Any single economy rate is a weighted average of three distinct baselines — and the bowler does not choose the weights. The captain does, the match situation does, the opposition's composition does. A bowler who bowls all four overs at the death at 9.5 is actually 0.7 better than his phase baseline. A bowler who bowls all four in the middle at 8.0 is actually 0.1 worse. The scorecard prints them on the same line. That single error sets the entire market's price structure.

I therefore maintain a 'phase-adjusted economy' — actual economy minus the league baseline for those specific overs. It is fixed at the moment the role is defined, never after seeing the outcome. My own rule: declare the role before you look at the result.

The third index is dot pressure. Counting dots is not enough; where the dot happened is the whole story. In my model's estimate, a dot ball in the middle overs raises wicket probability in the final five overs by roughly three percentage points. The mechanism is simple: the free-swing budget shrinks, and the batter is forced into risk earlier than he wanted. Those three points are written nowhere. The scorecard records a zero. I count the silence between the overs, not the boundary at the end of them. Across two hours and fifty-two minutes, nobody keeps the sharecropper's account of the balls that never reach the rope.

The gap between two markets grows from this soil. The logic of a Dhaka selection room is not the logic of a Kolkata auction room. In Dhaka, trust is built on consistency and length of spell; in the auction room, price is set by scarcity of role and famine of powerplay overs. The same cutter is in contention for a place because of arithmetic, and invisible because of the absence of highlights. The stadium was empty; the numbers were not — in those post-pandemic series the stands held nobody, yet the dot-ball ledger kept running exactly as before.

The problem is that the ledger does not always hold. I separate two things carefully: correlation and causation. A bowler with a heavy middle-over dot count may simply be playing in a side with thin batting depth, or on a slow pitch, or against opponents whose set batters were already dismissed. Once you control for the batter's aggressive role, a short boundary on one side, and roving fielders, half the significance of that dot count may not survive.

There is another side that keeps me cold-headed. A death bowler's economy is poor because he buys wickets with cash. Miss a yorker and it is six; a career-damaging figure is written in that phase. A captain takes that risk deliberately, because a wicket stops the match. Everything argued above is equally true in reverse.

So I keep a standing base rate for my own overrides: before contradicting consensus, I need a phase-adjusted gap of 0.4 runs, otherwise I grade my own call and last year's calls in public — win or lose. Because the analyst who is always 'actually' correcting the room is not a statistician; he is simply a new eye test wearing a table.

There is one more trap. The finer you split roles, the more players appear as 'bargains the market missed'. I cap the number of custom roles per analysis, and each must be written before outcomes are seen. If that arbitrage never closes, the role was the artefact, not the market.

My pre-registered call for this moment, timestamped today: I will check data from six teams over the past month, mainly the phase-adjusted economy and dot pressure of the bowler with the 4-0-22-0 figures. If his phase-adjusted economy is at least 0.6 better than the league average, and that holds up in review, then at the next auction at least one franchise will bid above his base price. If not, my notebook will state plainly: this article was wrong — and the error will stay in the ledger too.