The Empty File: When Cricket's Data Stops Speaking

A record opened on my screen last night. N/A where the title should have been, N/A where the source should have been, blank space where the one-sentence summary should have been. Twelve of thirteen fields held nothing at all; the thirteenth held two words — cricket_world. This was not a match report. It was the hollow frame of the report that was supposed to explain a match.

In March 2026, at Eden Gardens, I saw the mirror image of this. The Wills World Cup semi-final: Sri Lanka 251 for 8, India 120 for 8 in 34.1 overs, then fire in the stands, then the match awarded to Sri Lanka. Everyone in the press box wrote riot colour. I put thirty-seven completed scorecards in front of me and counted: between overs 18 and 34, India's scoring rate had fallen from 5.1 to 3.4. The collapse was legible before it was visible. That night the data existed and the verdict was missing. Last night it ran the other way — it was not the verdict that disappeared, it was the input. Absence is also a data point; you just have to change the frame you read it in.

Now the machinery. The first stage of the pipeline that produced this record does exactly one job: pull atomic facts out of the article body. One fact, one sentence, one verifiable unit — what the trade calls an information point. Everything else is joined onto those units: which player, which team, which format, which competition, how reliable the source, how fresh the news. Lose any one of them and the rest becomes scaffolding around nothing.

The format tag is the most brittle link in that chain. A Test batting average, an ODI average and a T20 strike rate are not comparable instruments. An all-rounder's economy rate across two formats reads like two different sports. If the record will not say whether this is a Test or a T20, every conclusion stacked on top of it is paper.

Which is why, in November 2026, when Mohammad Azharuddin was banned — after the Cronje story broke in April — I ran my 2026-2026 odds file backwards. Eleven one-day internationals surfaced in which the pre-toss price had moved more than three standard deviations from my model with no team-news explanation on the record. Eleven. I sent that number to two editors and one trader. Nobody printed it. The trader hired me. The ledger never lies, but it often whispers.

Last night's empty file is that same whisper, coming from the opposite direction. Twelve of thirteen fields read N/A. The entities field instructed the analyst to 'identify from the information points above' — and there were no information points above. That is the schema's structural flaw: it assumes an input arrived, and when none does, it quietly hands back blank cells. The domain label was present, but it read cricket_world where the specification called for 'Cricket'. A one-word mismatch looks trivial. It isn't. It says the article body never reached the classifier, or the parse zeroed it out, or language detection stripped the text before extraction. Any of those produces this exact result.

An empty template is not the danger. The hurry to fill it is. Someone inserts a player's name because that name was circulating all week. Someone adds a team's numbers because the numbers are remembered. Someone writes a 'result' because the result cell cannot be left blank. That is the only way twelve N/As get filled, and it is the only way the whole chain gets poisoned.

One more thing belongs on the record, because the analytics brigade is now inside the dressing room as well as the press box. Seam position, a batsman's body language, a captain's field settings — none of these appear in a spreadsheet column. And when a model goes quiet on a match, the first move is not to bolt on a second model. The first move is to watch twenty overs with the file closed. If the eye and the data disagree, report the disagreement rather than resolving it silently.

I reopened the anomaly file last night. The pattern had changed. The anomaly used to live in the numbers. Now the numbers are gone altogether. That is the larger event. A wrong figure argues with you. An empty cell does not argue — it invites you to fill it.

I went back to the source: every ticket, every note, every error. There was nothing in this record worth completing. It still did three useful things. It declared its own incapacity in plain terms — 'insufficient information, cannot assess' — which is rare honesty in this industry. It specified what should have arrived: a title, a source, a publication date, and at least one information point carrying fact, actor, format and competition type. And it flagged that if this record travelled onward without an ANALYSIS_BLOCKED marker, the reader would take it as 'no risks found' rather than 'no risks assessable'.

That, I think, is the real story, and it should be said plainly. The report that stopped at N/A is more honest than the hundred reports that put a name in the same place. In these transfer-window weeks we are surrounded by cricket's most abundant and least verified data: rumour. Every rumour is an empty template somebody has already filled — with a contract date, with an agent's name, with nothing but hope. The empty file is that template in its pure, unfilled state.

The counter-intuitive reading is that a null result is not the system failing; it is the system's healthiest component. It knows its own limit. A system that does not know its limit will publish a wrong ranking for a decade and nobody will catch it. After thirty years I still distrust the obvious odds, and now I distrust the obviously complete file as well. A filled cell means somebody decided something. An empty cell means nobody did — and that is a decision too.

The work this file leaves behind is a two-question filter. Next time you read a match claim or a signing report, ask: which format, and which source. If neither can be answered, the remaining numbers are decoration. In thirty years I have kept one standing rule: when the obvious odds turn out to be right, write that column too. The file must be allowed to close in the boring direction.

The next cycle's signal sits in three lines: how many information points each article actually carries, whether the domain label returns to 'Cricket', and whether every fact arrives with its format tag attached. If those three hold, analysis can begin. If they don't, then the next time an odds line or a rumour reaches your phone, one question remains — how many facts were in the file behind it? And if the answer is zero, what exactly are you pricing?