The Empty Cell and the Filled Illusion: When Cricket Analytics Confuses 'No Risk' With 'No Data'

In the winter of 2026 I hand-coded all 32 matches of the Tamil Nadu Premier League off the broadcast feed — 4,112 deliveries, each tagged with release height, revolutions and pitch zone. That database said Chepauk's first-innings par was 148, not the 165 sides kept chasing, and that 58 per cent of wickets between overs seven and fifteen fell to balls on a fifth-stump line. Chepauk Super Gillies won the title; nobody published the number that explained why. I did, in 900 words, for a site that paid ₹1,500.

Last week another such document landed in front of me. This time the missing thing was not a set of overs; it was an entire match. The first stage of an analysis chain came back with title N/A, source N/A, type unclassified, information points an empty list, core viewpoints empty. A payload carrying not one factual cricket entity: no format, no match, no player, no team, no league, no governance decision.

The most important sentence in that report was not that nothing was found. It was that the report refused to convert 'nothing found' into 'no risk identified'. Every row of the risk matrix read N/A, and one row was appended beneath them all — analytical risk: high. Because an empty payload is not a clean bill of health; it is a receipt for absent data. Confusing the two is the most expensive error available to the cricket industry.

Format first, verdict second

The first rule of cricket analysis is that Test, ODI, T20 and Hundred metrics do not transfer. A T20 strike rate of 150 and a Test strike rate of 50 describe two different competencies, sometimes in the same player. The ICC keeps three separate ranking tables for a reason, and blending them is not a stylistic choice but a technical error. Where a payload does not record format, every performance sentence written from it is inadmissible on day one.

This is not theoretical. On the 2026 England tour I logged 6,354 deliveries across five Tests — seam position, length band, lateral movement against the Dukes ball. In that ledger India's seamers went at 3.6 an over between overs 25 and 40, while James Anderson took 24 wickets at 14.5 and Sam Curran, Player of the Series, won matches in forty-minute bursts. All of it means something only because the format was fixed and it was Test cricket. The same deliveries in a T20 would make 3.6 an economy rate of destruction and 14.5 an average that ends a career.

The second rule is entity resolution. Where information points are empty, the entity field is empty too — and one empty list zeroes four dimensions at once. Player, team, league, narrative: all four are fed from that single list. The data is not gathered separately per dimension; there is one root, and the root is dry.

The third rule is money. Indian domestic coverage follows the money, and the money follows stars and finals, not middle overs and second spells. For years my seam database looked complete, until the missing overs started repeating. That was when absence stopped looking accidental and started looking structural.

Core: how an empty payload lies

What the report did was, technically, safe degradation. Eight dimensions, each in its mandated template, each cell reading 'insufficient information, cannot assess'. No estimates, no projections, no filler. Sporting, industry, timeliness and reference value each got one star, with the reason stated plainly: there is no subject to assess.

But the real question is what happens if that report travels downstream unlabelled. Four failure modes activate at once.

First, format contamination. When format is unrecorded, every downstream comparison is void, yet nothing in the report looks wrong. The benchmark set is missing, so any number can be compared with any other — and it usually gets compared with the most convenient one.

Second, automatic narrative entry. Where data goes quiet, the industry fills the space with story. 'Mystery spinner', 'the momentum shifted', 'the conditions changed' — all of it sits in the exact space where no measurement existed. Commentary names as mystery what it cannot measure. My whole fight with the spin-angle map was against this habit: by the thirty-second match the map had stopped agreeing with me, and I had to doubt my own reading before I could redraw it.

Third, an incentive pointing backwards. Our confidence systems reward completeness. A field filled with a plausible guess scores higher than an honest N/A. The system pays for false certainty and penalises the blank, which means the most honest analyst in the room files the weakest-looking report.

Fourth, silence rendered in green. An empty payload sits green on a dashboard, because raising a red flag requires a rule violation and there is no evidence of one. Absence of risk and absence of data get painted the same colour.

A borrowed frame is useful here, provided the borrowing is declared and its limits stated. Auction economics has repeated one principle for years: price is not international strength. Price is one number, cricketing capacity another, and the gap between them is the whole story of the auction. My import is the analogue: data availability is not analytical strength. The football-derived frame holds to the extent that both sports separate a number from a capacity. It breaks where cricket's discrete-event structure begins. Football is ninety continuous minutes; cricket resets after every over. A missing over in cricket invalidates the reading of one spell, not the entire match — which makes the cost of absence more local and more lethal.

Contrarian: absence is itself a selection effect

The easy conclusion is that data sometimes goes missing and a good system should tolerate it. The report did exactly that, honestly. But safe degradation carries a hidden complacency, and that complacency is the real blind spot.

When an instrument degrades safely on empty input, an organisation concludes that its instrument is conservative, and a conservative instrument is assumed to be an honest one. Safe degradation only proves the instrument is honest. It says nothing about whether the data underneath it is unbiased. Indian cricket's data is not unbiased. The overs that go missing do not go missing at random; they are precisely the overs that decide matches. Overs seven to fifteen, second spells, death bowling under a wet ball — this is the segment that gets cut first when coverage budgets tighten.

Which means the empty payload is not a neutral event. It is the end product of the same selection mechanism that stripped the middle overs out of the original coverage. In stage one, overs disappeared from the match; in stage two, the match disappeared from the overs; in stage three, somebody read the result as 'no risk identified'. One disease, three stages.

And here the more uncomfortable point surfaces: in cricket the fastest-spreading information is not what has been measured but what could not be measured. We cannot let go of the phrase 'spin-friendly' because it has no coordinates, and what has no coordinates gets turned into a story. That is why I stopped calling pitches spin-friendly and started describing release angle, drift and the batter's reach envelope. The day the map stops predicting, I treat it as a finding rather than an instrument — a failed model is still a result, because it is still a fact.

Takeaway: what to check in the next report

The report closed by listing six minimum inputs without which no further analysis is possible after re-extraction: format, nature of the event, at least one named entity, at least three discrete information points, a temporal anchor, and source identity. I would add a seventh, drawn from my own filing history — a pre-declared falsification fixture. Name in advance which match, which spell, which number would break your model. An instrument that does not know its own failure in advance cannot read its own success either.

Five signals are now in my sheet: whether re-extraction succeeded, whether a format tag appeared, whether an entity was recovered, whether the temporal anchor clarified, and whether source identity returned. Three greens and the analysis can run. Zero greens and the honest title of the document is not 'analysis' but 'data gap'.

So the next time a report says no risk was identified, one question has to be asked, and it is a methodological question rather than a polite one: was there no risk, or was there no data with which to measure risk? Until that distinction sits in the headline, every cricket dashboard will keep mistaking silence for cleanliness.