When Labels Lie: Football and the Cost of Misplacing a Position

2:47 a.m. in São Paulo. I sat in front of the screen with my third cup of coffee, waiting for the system to push the new piece for my weekend column. The file arrived, clearly labeled: football domain. I opened it. Inside were seven million mobile phone lines in Mexico waiting to be cut off. Not a single club. Not a single player. Only a telecommunications regulator, a deadline of September 15, and a list of phone numbers sorted by last digit, ending with groups of numbers ending in 0, 1, then 2. I closed the file, opened it again, and found the same thing. A document about mobile subscriber identity registration had slipped through the exact gap that was supposed to be reserved for tactics.

That night I did not sleep. Not because I pitied those millions of subscribers — that is another industry's business. I did not sleep because I recognized something familiar enough to make my skin cold: our football is slowly dying from being mislabeled.

Nearly a decade watching this industry, I have learned that labeling is the most powerful act almost no one notices. A player called a winger will be judged by the number of crosses. A midfielder called a number 8 will be scrutinized by his sideways passes. A document called football will be handed to a football analyst. The label goes first, the truth follows — or rather, the truth gets locked inside that label.

That Mexican file was a reminder. The machine labeled it, done. No one checked. Then it drifted into exactly the room it did not belong in. In the era of automated data, we are building an entire analytics operation on labels generated by machines, and almost no one opens the box to see what is actually inside. Seven million lines. A citizenship ID code. A September 15 deadline. Not a single word of football. And yet the label still read football, and had I not bothered to open the file, I would have sat there analyzing it like a quarter-final.

Football learned how to label long before machine learning existed.

My writing career began with one mislabeling. In 2026, aged seventeen, I wrote a long piece about Vinícius Júnior, then sixteen, playing for Flamengo. The whole football world called him a speedy winger. A tidy label, easy to understand, easy to sell. But the data I gathered from under-20 matches told a different story: twelve successful dribbles per game, eight chances created for teammates. That is the profile of an inverted-left number 10, a game organizer, not a wing-running machine.

I wrote that he had to play as a false 10. People laughed. Daring to compare him to Ronaldinho, they said. Three years later, Real Madrid used him in exactly that role. And I learned my first lesson: most football arguments, at bottom, are arguments about labels.

Vinícius is not rebelling; he is re-reading the tactical map. A mislabeled player has two choices: accept the label and play worse inside its limits, or wander out of the box to find his true place. People call that a reaction, an attitude, an ego. I call it the act of re-reading the map when the old map is wrong.

Now look again at that Mexican file. The same mechanism. A dataset placed in the wrong box. It will be handed to the right person, that person will analyze it enthusiastically, and no one will ever discover that the box was mislabeled from the start. In modern football, dozens of processes run exactly like that every single day.

The biggest enemy of analysis is not bad data, but a bad label attached to good data.

Those seven million lines, had I fed them to a model, would have returned very confident conclusions. It would speak of subscriber growth, of disconnection risk, of compliance rates. All of it fluent, all of it numbered, all of it sounding true. That is the most dangerous part: a wrong label does not produce an obvious error. It produces confidence. It turns a telecom document into a smooth tactical report with not a single crack.

I have seen this on the pitch. Last season I tracked a match in which the home side was described as a high-pressing team. The chart showed them pressing. The numbers showed them pressing. But when I rewatched every sequence, almost every pressing situation came after the ball had already been passed away — a chase-pressing, not a block-pressing. The label said one thing, the footage said another. And because everyone trusted the label, no one bothered to rewatch the footage.

That is why I keep an iron rule in this trade: whenever a number shocks, I seek at least two independent sources before writing. Not to be sure I am right. But to be sure I am not analyzing a Mexican file disguised as football.

Based on my experience watching matches, there is a tell-tale sign of a wrong label that a machine never sees. It is excessive fluency. When an analysis is too tidy, when every piece fits perfectly, when not a single detail deviates from the story — that is usually the moment someone labeled first and then stuffed the data in to fit. Real football always has noise. A defender who plays well in the first half and vanishes in the second. A striker who scores and then disappears for three matches. That noise is the truth leaking out of the box.

And that Mexican file has no noise. It has only a wrong label, clean and confident.

So where could I be wrong in this whole argument?

There is one possibility I must admit honestly: maybe the label was not wrong, but I misread it. Perhaps the system never said this document is about football, only said process this document through the football pipeline. Those are two different things. If so, the fault is mine — a man who read an operational label too quickly and mistook it for a statement about content.

And there is a second, more uncomfortable possibility: maybe labels do not matter that much. Football has survived decades of wrong labels — misunderstood players, misnamed systems — and still produced beautiful matches. If a wrong label cannot kill football, then my worry is only the worry of a writer inflating the importance of his own trade.

I accept both possibilities. I simply do not accept silence in front of a label no one checks.

The number 10 is not dead, it has just learned to run faster. And perhaps it is now the labelers' turn to learn how to run — to run backward, to check whether the box they just closed truly holds what they think it holds.

If this trend continues, within two seasons football analytics rooms will no longer argue about tactics. They will argue about labels. The club that labels its players more honestly will go further — not because it has better data, but because it does not fool itself with a wrong name. An academy brave enough to call a playmaker a playmaker, instead of calling him a winger to sell him sooner, will raise real number 10s instead of their shadows.

As for me, that night, I finally slept. After closing the Mexican file, opening an old Flamengo match, and reminding myself of a line I know by heart: open your eyes before you open your mouth. That is the whole job. And I believe it always was the whole job.