Mislabeled on the Pitch: How Dirty Data Slips Into Vietnam's Football Analysis Rooms

On January 5, 2026, at Rajamangala, Nguyen Xuan Son lay on the grass. He left the pitch on a stretcher, and Vietnam still won the ASEAN Cup 5-3 on aggregate over two legs. His seven goals in that tournament were the most quoted figure of the week.

Then, within roughly forty-eight hours of the final whistle, I ran into a different kind of data, far quieter. One post called that final a "V.League derby". One page automatically filed it under "international friendly". One stat sheet assigned Xuan Son's name to a match he never set foot in.

Nobody invented a new match. A label was simply stuck on wrong, and from that label, everything downstream went wrong automatically.

That was when I opened the spreadsheet.

Three harmless data points stitched together become a money-flow map leading to a village with no football pitch. I still use that line when I talk to first-year students in Lyon, and it holds exactly as well when the subject is Vietnamese football.

An ecosystem running on labels

My trade is reading books and cross-checking data. I was born in Brazil, I work in France, but for ten years I have followed Southeast Asian football as a verifier rather than a commentator. What I found in the Vietnamese market is an ecosystem that runs on labels more than on numbers.

There are three layers. The first is raw data: a V.League 1 organised around fourteen clubs, federations, match-statistics providers. The second is the press and aggregator layer, where an event gets tagged, categorised and pushed into a section. The third is social media and short-form content, where labels get reused with almost nobody checking them again.

The problem is that the third layer flows back into the first. A wrong label at layer three, after a few cycles of sharing, becomes a definition. And once it is a definition, nobody asks whether it is true.

I have sat long enough in scouting rooms to know that people rarely re-check the label. They check the number. And a number always looks very solid, even when it was born from a wrong label.

Unpacking four layers of error

The first layer of error is polysemy. In Vietnamese, "gala", "elimination", "host", "derby" and "final" are words that live in several worlds at once. An awards night, a television knockout round, a genuine local derby: all can carry the same tag. An automated classifier cannot tell context apart, because it only sees the word.

The second layer is time. A clip shot in 2026, reposted in 2026 with a new description, instantly becomes "this season's data". I once spent three weeks cross-checking a video described as a foul in a V.League match, only to find the goalposts in the clip were painted a different colour from the recorded season. The sponsor logo on the chest belonged to a deal that had expired two years earlier. A small detail like that is enough to break the whole chain of reasoning behind it.

The third layer is identity. Vietnamese football has many players sharing family names, and naturalised players whose names sit close together in international databases. When an overseas site assigns one man's minutes to another, the error does not vanish. It travels into scouting files, into valuations, and eventually into the decision of a club in another country.

The fourth layer, and the hardest to talk about, is that labels are stuck on for money. Content tagged "football" enters a completely different advertising market than content tagged "entertainment". For the same audience size, commercial value differs several times over. When the reward sits in the label rather than the content, the system will produce labels.

A label is not a description. A label is an economic decision, and it is made before anyone verifies the content.

The consequences arrive in a very dry manner. A single wrong label is harmless. Three wrong labels stitched together become a pattern. And when the pattern gets large enough, it starts producing what analysts call a false positive inside the data pipeline: a record carrying a football label with not one minute of football inside it.

I have met exactly that case. A file pushed into the football category, complete with familiar keywords: final, elimination, host, lineup. But when opened, the whole content revolved around an argument between television presenters. No club. No player. No score. Not a single data column.

The correct handling in that situation is simple and very hard: write into the conclusion field that there is not enough information to assess. Do not speculate. Do not try to give it any football meaning. Because once you start inferring from material belonging to the wrong domain, you are no longer analysing. You are writing fiction.

People call me a sceptic; I call myself someone who reads the books behind the pitch.

Rajamangala and the data gap

Back to Rajamangala. Xuan Son's injury was clean data medically and very dirty data in media terms. Within three days I counted a stream of items assigning him different minutes totals for the same tournament, differing by several hundred minutes. Nobody lied on purpose. They simply copied from a source that had been mislabelled from the start.

What stands out is that fixture density, which I still regard as the single biggest cause of injury, barely appeared in those reports. Nguyen Quang Hai and Nguyen Tien Linh also sit in the group of players shuttling non-stop between club and national team, and their stories were handled in exactly the same way. Attention went to the moment of contact, to the image of the stretcher, to emotion. The real question, how many matches a player had played in how many days before the bone broke, lay scattered across a few data tables that nobody stitched together.

The lesson I took from the 2026 World Cup: referees can read a spreadsheet too. That year I sat down for a month to hand-log more than a thousand decisions, and what I learned was not that referees favour anyone, but that public data always has a gap, and that gap is usually filled with feeling.

In the V.League that gap is wider. A fourteen-club competition, not many matches, but the number of versions of each match online runs dozens of times higher than the number of real matches. Every goal exists alongside dozens of re-edited clips, recut, re-scored, re-labelled. The original is still there. But the original is not the most-watched version.

When VAR appeared in some matches, I gained an extra layer of cross-checking. I also gained an extra layer of noise, because each decision is recorded differently in each place, and nowhere records the reasoning behind it.

The transfer market never lies if you read the agent-fee column instead of the player-price column. The same principle applies to information: do not read the description, read the classification.

The counter-view: stop demanding perfect data

Here I have to argue against myself a little.

Mislabelling is not the biggest enemy of Vietnamese football. The bigger enemy is the habit of refusing to say "I don't know". In my trade, the hardest sentence to write is not the accusation. The hardest sentence to write is the blank one.

If you demand perfect labels, you get a system that dares not label anything new. A strange clip, a young player never recorded in a lower division, a match at a ground with no camera: all get pushed out because they do not match an existing category. Loose labels are the price of discovering things that do not yet have names.

The reasonable part of wrong labels sits there: they open a path to the unknown. The dangerous part sits there too: they never announce that they are wrong.

I do not believe in perfectly clean football data. I only believe in data that declares its own blind spots.

And I do not believe every mislabel is a conspiracy. Most are laziness, deadline pressure, copy-and-paste habit. Only a small share is deliberate. But that small share is enough to ruin the larger part, because it picks exactly the places nobody checks.

Sports culture looks finest from the stands; it looks foulest from the accounts office. And between those two places sits the data room, unloved, unphotographed, and decisive for most of what you will believe the next morning.

This holds for more than one league or one country. It holds for every market where the speed of news outruns the speed of verification.

What I leave behind

If you follow Vietnamese football and next week you meet a description that sounds entirely reasonable, do one thing: check the label before you check the number.

The question I leave to those building tools for millions of domestic viewers: are you building a system to classify football, or a system to sell football labels? The two look so alike that they are hard to tell apart, until a file with no football minutes in it slips into the analysis room and nobody notices in time.

I still keep the spreadsheet. The first column is always the label. The second column is the number.