Nine chapters, zero facts: how a broken football data pipeline reaches the reader

In 2026, during a research project on how empty stadiums affected match results in the V-League, I almost published a conclusion taken from a spreadsheet with one blank column. Eighteen rounds, complete data entry, formulas running correctly, clean charts. Home-team win rate fell from 42% to 31%. That figure was strong enough to carry a headline. But when I reopened every row, the notes column recording starting line-ups — the column that let me strip out the human factor from the home-advantage variable — was completely empty. Not empty because there was nothing to record. Empty because I had forgotten to type it in.

The frightening part was not that I nearly got it wrong. It was that the spreadsheet still looked flawless. No red exclamation mark, no warning, only a white silence sitting among hundreds of cells of numbers. Had I not reopened it myself, that conclusion would have walked out of the door.

Recently I received a document very much like it, only larger. A nine-chapter football report, more than forty rows of data, neatly ruled tables, a high-risk section, a recommendations section, and even a glossary of professional terms at the end. The only field with any content was a label: football. The original headline was blank. The outlet name was blank. Publication date was not assessed. The list of information points was entirely empty.

A report that thick, and not a single fact to verify.

Football analysis today runs on two layers. The first reads the source article and breaks it into discrete information points: who, did what, when, where, according to which source. The second layer — where I work — takes each of those points as a foundation for tactical, financial, disciplinary and media analysis. The principle is simple: every conclusion at layer two must trace back to a specific information point at layer one. No points, no conclusions.

During the transfer window, pressure on layer one peaks. Rumours about transfer fees, release clauses, new wage bills, agent activity — any one of them can generate hundreds of thousands of reads within hours. When speed is the criterion, verification gets treated as a cost. I have watched transfer stories spread across every outlet on the strength of one phrase — according to a source close to the deal — with no name, no age, no job title.

But rumours are not the trap. Readers know to doubt rumours. The trap is the report that looks absolutely precise, presented as a technical product, while the material inside has long since rotted.

In the data industry this phenomenon has a name: null propagation. An empty record at the intake stage is not corrected, not blocked, but passed along. The next stage has no way of knowing the previous stage failed, so it treats the record as established fact and produces its own output. The result is a document with a complete shape, complete structure, and no value.

An empty data point is not a neutral data point. It is contagious data. A blank cell does not announce itself as blank after passing through three processing stages; by the time a reader sees it, it is wearing the clothes of a statistic.

At the technical layer, prevention is not complicated. Put a validation gate immediately before the record leaves intake: if the information-point list is empty, or the title and source are both blank, the system must return an ingestion-failed status rather than pass it forward. A sensible minimum threshold — title, source, and at least three source-attributed information points — blocks most cases.

What stands out is that the report I received was not messy. It was written in correct analytical register. Comparison tables still had rows and columns. The risk section still sorted items by level. The recommendations section still proposed concrete action, including halting distribution and rerunning the intake stage. In other words, the system knew it had failed. It simply had no way of saying so where the reader would look.

I first met this exact logic in 2026, working as a statistics volunteer at a national U19 match. My job was to log 47 foul situations and 12 offsides. In the 78th minute I saw a collision inside the penalty area that the referee waved away, immediately before a controversial goal. I cross-checked it against Law 12 of the IFAB Laws of the Game, built a comparison table, and sent it to the organisers. No reply.

It took me a week to understand: that silence was itself a data point. Not a fact about the laws, but about the process. A file with no recipient is not a file. I learned from U19 football that mistakes are less frightening than nobody measuring them.

In 2026, drawing on my experience tracking matches, I built my own spreadsheet covering all 64 World Cup games: completed passes, possession share, sprint counts. The champions ranked only seventh for possession. I published the analysis and was told I did not understand football. I rewatched the full footage of ten of their matches and rechecked every number. Not one figure was wrong. What was wrong was that I had not placed them in the context of the match.

That lesson applies directly here. Football does not change because you look at it more closely. Football changes because you look at it more correctly.

At the professional layer the problem is more serious than at the technical one. A validation gate can be programmed in a few hours. But the habit of reading a report by its form takes years to fix. When a document has nine chapters, eight tables, a risk section and a recommendations section, readers tend to judge it by density of words rather than density of facts. Thicker means more trustworthy — a bias with no basis, but one that performs very well.

At the market layer, the consequences are measurable in money. A club extends a contract on the strength of a fitness report with no GPS data. An investor values a player using a statistics table missing the minutes-played column. An outlet publishes a transfer story sourced to a nameless contact. In all three cases the input was empty at precisely the point that mattered most, and the output looked extremely full.

Most people's first instinct on seeing a blank table is to fill it in. Estimate, extrapolate, use a comparable sample, apply professional experience. This is the moment an empty analysis becomes a wrong analysis — and the second kind is far harder to detect, because there are no blank cells left to trace.

The rule I set myself from the 2026 season: never use a single season to assert a trend. Eighteen rounds gave me a home win rate falling from 42% to 31%. At a glance, a major finding. But the sample was small and squad effects dominated. My supervisor at the time asked me to compare against five previous seasons of data. I did, and the conclusion dissolved. I do not believe in luck. I believe in the number that repeats a hundred times.

Empty stadiums taught me this: noise never scores.

There is another side to this that is rarely discussed: the values that matter most often sit in no cell at all. A midfielder like Kanté leaves no obvious trace on the record sheet — few goals, few assists — yet appears in every clip closing exactly the gap the opponent wanted to attack. Read the numbers alone and you conclude he is ordinary. Read only by feeling and you measure nothing. Every slow-motion replay carries its own truth. My job is to find the truth that cannot be argued.

And here is the final counter-intuitive point: the most dangerous report is not the one missing data, but the one full of data nobody verified. The empty one indicts itself with its blank cells. The full one does not.

In a transfer window, when every account has a story, every story has a number, and every number looks plausible, the value of a data analyst is not in adding more numbers. It is in daring to say this cell is empty, and daring to stop the line.

A referee's decision is only the end point. The real journey lives in every camera angle.

Football analysis would advance far faster if every newsroom ran one simple gate before publishing: title, source, date, and at least three traceable facts. No artificial intelligence required, no complex model. Just one line of code that says the conditions are not met.

That nine-chapter report was never published in the end. It was held back exactly where it should have been held back: before the reader's eyes.

Next time a transfer story reaches you complete with fee, clauses, contract length and a quote from the agent — will you ask yourself who verified the first cell?