412 Rumours, 64 Matches, 11 Weeks: How Cricket's Data Audit Fails

On 15 July 2026, forty-eight hours after the World Cup final in Russia, I closed a spreadsheet. Sixty-four matches. I had logged PPDA and expected goals for every one of them myself, at two in the morning, after each fixture. I did not write a single word about the outcome after the final — I only checked the numbers twice. I recalculated Germany's group stage: 5.6 xG generated, two goals scored, four conceded. Croatia covered 1,116 kilometres across seven matches, the highest in the tournament. Those numbers were my only witnesses. Not the headlines. Not the highlight reels. Just timestamps and raw event feeds.

Today I want to talk about that method, but not about the cricket field — about cricket's information discipline. Because a large share of the reports landing on my desk in recent months look right. The structure is right. There are numbers in the headings. And the inside is empty. That is the most dangerous kind of error.

A report with no format is not a report. It is a blank form.

Let me be precise about what happened. A report arrived from an analysis pipeline. Eight chapters, tables in every chapter, rows in every table, an assessment in every row. Magnificent to look at. But on the first page there was a line: the information-points field was empty. No title. No source. No player name. No team name. No venue. No format. Just one tag — cricket_world.

In that condition an honest analyst has exactly one job: write insufficient information in every cell. Do not fill cells with imagination. Because if the format cannot be identified, no phase-based commentary is admissible. Powerplay, middle overs, death overs — those words sound professional, but if you do not know which format you are describing, they are meaningless. Judging a T20 batter by a Test average is the oldest error in cricket, and it has now migrated into data reports.

The cascade from a zero information-point load is a chain: no name, so no role; no role, so no idea which metric even matters.

This is the real problem. The pipeline was instructed to extract entities dependent on the information points. Empty information points mean empty entities. The instruction for grading source quality was to judge from the source fields of the information points. No information points means source quality is ungradeable. Time sensitivity, same. One failure became three, because all three were anchored in the same place.

I saw another version of this failure in January 2026. I logged 412 transfer rumours published about Championship clubs in the UK press during the winter window. Only 47 completed — an 11.4 per cent hit rate. I built a four-tier source spreadsheet and treated each journalist's record as data rather than reputation. That thread reached 300,000 impressions, and three agents asked me to stop.

That experience taught me one rule: a claim is a claim only when it carries a timestamp and a source tier. In a report with empty information points, both are missing.

So is the report entirely false? No. One part of it is true, and that part is the most useful thing in it — the pipeline's self-diagnosis. The report itself admits that an empty object passed schema validation and reached the downstream stage. That is the actual crisis. The report contains not one valid conclusion about any cricket team, player, board or market. But on the research process it has produced a valid, serious and cheaply fixable conclusion.

A neat but empty report is more dangerous than a failed one, because the user mistakes structural completeness for analytical depth.

This is my central worry. Football or cricket, in any data culture we mistake structure for quality. Eight chapters do not mean eight kinds of analysis. Eight tables do not mean eight kinds of evidence. I have personally corrected myself three times on this — three claims I made on small samples in 2026 I publicly retracted myself. No reader caught them. I caught them.

The same principle applies here. A report with no format, no venue, no player cannot say no risk. It must say not assessed. The distance between those two sentences is enormous. One is a clean bill of health. The other is an acknowledgement of an incomplete audit.

I recalculated those sixty-four World Cup matches because my first count was wrong. In eleven weeks of furlough I built a 4,000-match database because even knowing the phone would not ring, an empty calendar still accumulates data. I did not write one word about the empty-stadium effect until 200 matches had been played — the home-win rate had fallen from 43 per cent to 21 per cent, but the sample was small, so I waited.

Now the question is not about cricket. It is about cricket journalism. How many reports do we publish that look complete but cannot withstand a single question? And how many do we withhold because they are honestly incomplete? In my accounting, the second group matters more.

If a reader opens a report and finds insufficient information written in every cell, the author is honest. If every cell carries a confident assessment but no format is named anywhere, look for the timestamp. If you cannot find one, understand this — those numbers are nobody's witness. They are somebody's guess.

The first number I checked was not the fee. It was the timestamp. This report does not have one. So its value is not zero — its value is a warning.