A 'Football' Label on a File With No Football: When the Sport's Classification Systems Fail

The file arrived at 6:40 a.m. Liverpool time, with a single word on it: Football. I opened it, read all thirty-five information points, then sat still for a long while. There was no club in it. No player, no match, no tactical shape, not a line of xG. There was only Intocable, Calibre 50, Paty Cantú, Kabah; only Zócalo, Grito de Independencia, and a concert schedule in Mexico City on September 15. Thirty-five data points, and not one of them belonged to football.

My job is transfer market administration. Every day I read hundreds of files like that one. Most of them arrive with a label, and most of the time the label is right. But that morning reminded me of something I learned long ago, before I ever knew what xG or PPDA meant: a label stuck onto a thing is never the thing itself.

I entered the trade in 2026, in the sports department of a television station, and for nearly thirty-five years my work has been mostly about classification systems. In modern football, almost everything passes through a label before it passes through a human eye. An event gets tagged. A player gets tagged by position. A match gets tagged by competition. A news item gets tagged by topic. The label is the first thing we see, and very often the only thing we read.

Picture a scout in the Premier League. He cannot watch four hundred players. He watches the forty who survive the filter. And the filter, in the end, is a string of labels: position, age, minutes played, league tier, contract length. If one label is wrong, the shortlist is wrong. If the shortlist is wrong, he still goes to watch exactly forty players, but the wrong forty, and the conclusion he reaches afterwards still looks scientific, still looks well-founded.

I have watched this drag on for years. A winger was tagged as a full-back in a major database; for two transfer windows, clubs searching for left-backs under 23 simply passed him over. A young talent was still marked U21 at twenty-two and a half. Old labels do not update themselves. And when football has become an industry where a single buying decision is worth tens of millions of pounds, an old label can cost more than an injury.

That is why the morning with the Mexico City file made me stop. Most classification errors in football are partial: right topic, wrong detail. That file was a total error: the label said football, the content was a music festival. It is the cleanest case for dissecting the mechanism, because when something is one hundred percent wrong, you cannot soothe yourself with the phrase minor oversight.

I have a habit: whenever my own classification systems fail, I go looking for the same fault in myself before blaming the machine. And I found it, more often than I care to admit.

In 2026, I built a homemade xG model to analyse all sixty-four matches at the World Cup in Russia. I predicted France would win from the group stage, based on an average chance creation of 2.4 xG per match. I said Croatia had low xG but good luck. I was right about the champion, and badly wrong about the rest. xG is a revolution, but every revolution needs time before people accept it — and I too needed time to accept that my model had a hole: it ignored corners. A variable that simply did not exist inside the label live-ball situations that I had assigned to it.

I remember locking myself in a library for two weeks to rewatch the data. The person who is right before his time always pays in loneliness. But the worst loneliness comes when you realise you were right about the conclusion and wrong about the method.

Two years later came the pandemic. In March 2026, football stopped. Liverpool were twenty-five points clear of Man City and almost certain champions, then the season was suspended. I wrote three drafts and deleted all three, because I no longer trusted my own spreadsheet. When football returned to empty stadiums in June, the data forced me to look straight at it: the home win rate fell from 46 percent to 39 percent. No label in my system said empty stadium. An empty stadium does not distort the data, but it makes the truth feel hollow, and I had to sit alone for weeks to redefine my model.

There is a counter-example, to show where a correct label is built from. In 2026-18, Juergen Klopp's Liverpool finished in the Top 4 with 78 points through a frantic pressing style, with Mohamed Salah, Sadio Mané and Roberto Firmino ahead of it. I calculated their average PPDA at 8.2, the lowest in the league, while Jose Mourinho's Manchester United sat at 15.7. Based on my experience of watching matches at Anfield that season, I wrote a long piece on gegenpressing and was criticised for being too mechanical. I once stood before a data table and felt I was watching a match retold in another language. Data whispers, and those who listen will hear a miracle — but only when the label attached to that data is not lying.

In the transfer market, the label costs far more. When Liverpool paid around 85 million pounds for Darwin Núñez in the summer of 2026, most of the debate was not about the player but about the label stuck onto him: number nine, centre-forward, Sadio Mané's replacement. I have said many times that the youth price bubble is bursting, and that one hundred million euros for a player who has not yet played fifty top-flight matches is a naked gamble. But here is what few say: that gamble is usually placed on a label, not on a match someone sat through for a full ninety minutes. In a world of seasons that never end, the awakened can only rely on their own spreadsheet — and must check what name that spreadsheet is calling the player.

The easiest thing to think, reading about the Mexico City file, is to blame the algorithm. I do not go that way. A label is not a fact; it is a claim — and a claim always needs someone to verify it. The fault here lies not in the machine that applied the label, but in a process that lets a label go straight into a decision with nobody opening the content to read.

In football, we are already familiar with a similar mechanism called VAR. Video referees are taught to intervene only for a clear and obvious error. But that very phrase is a vague clause: clear to whom, obvious by what standard? The space for subjective judgement in VAR is larger than people think. Likewise, the word Football sounds decisive, but it is only decisive until someone opens the file. The fault is not in the letters on the label; it is in the fact that we stopped opening it to read.

And I have to add something uncomfortable: I myself have been a poor labelling machine. My 2026 World Cup model stamped enough data onto a set from which it had omitted corners. The arrogance of the person holding the numbers is no different from the arrogance of the person holding nothing at all.

In the next cycle, I think fewer clubs will ask what the label says and more will ask who opened the file to read it. It is the kind of improvement that makes no noise and earns no headline, but it determines exactly which sums a wrong label can steal in silence. For someone in transfers, my first task each morning is no longer to filter, but to open a few random files and read them — because next time, a lying label may not sit in Mexico City, but right inside a club's shortlist.