Blank Cells at the Death: The Numbers We Refuse to Count

Blank Cells at the Death: The Numbers We Refuse to Count

The fifth ball of the 18th over last night: the bowler went searching for a yorker and delivered a low full toss; the batter sent it over long-on. In my hand-coded sheet that cell still reads unfinished — ‘18.5: dot? boundary?’ Three years and I have not erased the question mark. None of the feeds circulating in the circuit have a column called ‘intent’, and that blank cell is exactly where the real story of the match gets stuck.

The scorecard will tell you the bowler’s death-over economy is 9.8, which sits comfortably inside any budget. In his last four matches of this tournament he bowled 96 balls between overs 17 and 20; 59 of them were length or low full toss, only 22 were yorkers. The 96 and the 59 are mine, measured — not handed to me by a feed. Put those two numbers side by side and you see a different bowler, and a different selection decision.

When the national shirt goes into a tournament, every failure becomes a national question overnight. Over the last two weeks, in three innings, the team has conceded 31 runs in the final five overs — but the 45,000 people in the stands and the thousand posts online cannot tell you how much of that 31 was bowling failure and how much was the size of the ground. Emotion wants a fast decision, and the tournament format punishes exactly there — nobody has done the squad-depth arithmetic.

My method is simple, and it forces humility. I opened a blank spreadsheet and let the Bangladesh Premier League teach me. Five columns per ball: over, bowler type, length, batter’s hand, venue. Then I label every number — measured, modelled, or guessed. Any column marked ‘guessed’ cannot father a thesis.

The xG model was crude, but the missing cells confessed more than the goals. In cricket that job belongs to the phase split: powerplay 1-6, middle 7-15, death 16-20. Across the 41 matches of this tournament I hand-coded, middle-overs dot-ball rate averages 38 per cent. That sounds impressive, doesn’t it?

The trouble is that 38 per cent says nothing on its own. When a leg-spinner bowls to a team’s number seven with 800 runs already behind the game and a deep field set, that dot is arithmetic, not pressure. The bowler who denies a set batter at the 18th over logs the same figure in the same column — and it is effectively a different sport. Dot-ball percentage and ‘runs saved’ are T20’s trap, the place where a measure of effort quietly becomes a measure of victory.

Football knows this trap well — distance covered and high-intensity sprints. A midfielder can run 12 kilometres and lose everything, and by the following week that becomes a headline about ‘tremendous work rate’. In T20 its equivalent is the dot ball. Three teams in this phase have bowled more dots than their opponents; two of them are already on the way out, because their dots came against batters who are happy to take them.

And who is actually filling these blank cells? The answer is uncomfortable. In domestic and associate-board matches the data is entered by the league’s own scorer, and his sheet does not even have a field for length. Wickets, runs, overs — nothing else is captured. What nobody collects, we later sell as ‘incredible form’. Even international feeds have only started separating yorkers from low full tosses in the last few years.

Venue and dew are the next layer. In the second innings, once the ball is wet, the grip goes, and a failed yorker becomes an environmental fact rather than a bowler’s fault. Compare boundary dimensions in Sylhet and Rangpur and the same length is four in one ground and a catch in the other. Yet no feed’s economy column carries a venue adjustment. A spreadsheet without venue, dew and innings state is not data — it is a story with numbers attached.

Now my model. Five inputs only: batter quality tier, bowler’s length mix, ground dimensions, dew, and innings state. I chose the weights by hand, so the error band drifts between six and ten per cent; across a 41-match sample, a single spell almost always lands inside that band. Hand-picked weights have one virtue — they do not claim to know the truth, only to declare what they left out.

Since Russia 2026 I watch Germany twice: once with eyes, once with PPDA. The press had drifted from 8.9 in qualifying to 12.6 at the finals, and the match looked like two different games. In cricket that two-track habit now lands on a single question: when I praise a death specialist, am I watching him, or watching the easy spells he is given to bowl?

This is where correlation and causation split. The man we call a death specialist may owe his tidy numbers to a simple fact — by the 16th over the opposition has usually lost its top order, and by the 19th it is the number seven or eight at the crease. The bowler who takes the 19th over against the top order spends the same spell and pays more for it. Not causation: sample selection.

Selection matters just as much. When a captain picks an extra bowler, it is often not about adding options; it is about avoiding blame — nobody wants the reputational risk of breaking the four-bowler structure, because failure carries his name. Batting depth then charges its bill at the 17th over, when the big shot comes from someone whose strike rate is 20 points below the set batter’s.

Workload is the harder question still. A quick returning from a knee ligament or shoulder injury takes longer in the head than in the body; six months of fearing the yorker cannot be undone by handing him the 19th over. That is not a match-up calculation, it is a risk calculation. Rushing it widens the crack, and a model like mine cannot measure it — nobody has built a column for courage yet.

Mustafizur Rahman’s cutter-yorker and Lasith Malinga’s death-over routine were the product of long habit, not one season’s discovery. Rashid Khan’s leg-spin shows how pressure is built in the middle overs, which should tell us the phase ledger does not begin at the 16th. So in the next round my first column will be a different number: death economy weighted by batter quality, with sample size and venue adjustment attached.

The teams that play the rest of this tournament have only one question to answer — who bowls your last five overs, and what number does that batter actually bat? The side that asks this before the toss does not fill the blank cells on the scorecard; it learns to read them. And silence is not zero; it is a new baseline with its own residuals.