Wrong Labels Break Analysis: Why Vietnamese Football Data Must Be Re-Audited at the Source

At 11:40 p.m., after the weekend's V.League round had wrapped up, I opened a file labelled "Football." Inside were Susana Zabaleta, a Mexican singer and actress, and Ricardo Pérez, a member of the comedy troupe La Cotorrisa. No team. No match. No player. Not a single second of football footage.

I am not telling this story for laughs. A wrong label at the collection layer does not stay at the collection layer — it runs down the entire processing chain behind it, and by the time it reaches the reader it is wearing the clean jersey of statistics. Vietnamese football is building its own data system on exactly that kind of pipeline.

"The louder the stands, the easier the truth hides."

Over the past decade, the way the football industry produces knowledge has changed at the root. Analysts used to sit through recordings and write every action into a notebook by hand. Today, most of what you read about V.League passes through three layers: automated collection, keyword-based labelling, and redistribution through aggregator sites. Every layer can fail, but only the first can be fixed. The later layers merely amplify the error.

In Vietnam, most sports newsrooms run on thin staffing and hourly publication pressure. Nobody has time to inspect each source file. A mislabelled "Football" tag will automatically generate dozens of articles, hundreds of tags, and eventually a cluster of data that sits quietly inside search systems, waiting to be cited again one day.

At the technical level, labelling is more subtle still. A player may be tagged as a "winger" while he actually plays as an inverted attacking midfielder. A counter-attack may be logged as a "wide attack" though the ball never travelled down the flank. A defensive metric may be expressed as PPDA but sampled from matches in which that team deliberately sat deep.

"I don't trust my eyes; I trust what my eyes cannot see."

I tracked Nguyễn Tuấn Anh through the 2026–2026 period by hand, the old-fashioned way. The first results made me think I had made an error: his distance covered fell 17 per cent, yet his accurate passes rose 23 per cent. Then I realised I had mislabelled him myself. I counted him as a box-to-box midfielder when in reality he played as a deep-lying playmaker. My label was wrong, not the player.

Most arguments about Vietnamese football data are really arguments about labels, not about numbers. People fight over whether a player runs a lot or a little, when the problem lies in where the system decided he plays. Change the label, keep the number, and the conclusion flips.

"The truth everyone believes" — that is the phrase I use most when talking about football data. Everyone believes data is neutral. Everyone believes a statistics table simply records what happened. But data does not record what happened; it records what the labeller decided had happened.

In V.League the problem is worse because there is no common standard. Clubs use different providers and different definitions of a duel, a key pass, a shot. Comparing two players at two different clubs then becomes almost meaningless. You are comparing two labels, not two people.

That flow runs straight into the transfer market. A young player with good numbers under system A gets priced by system A. The buying club uses system B, and discovers he cannot play the role the data table promised. The player gets blamed. In reality the fault sits in that first line of labelling, dashed off in an afternoon two years earlier.

"The man on the bench sees most clearly who is acting."

I once sat beside a fitness coach at a mid-table club. He said something I have never forgotten: his team's data was only right about sixty per cent of the time, and he knew exactly which sixty per cent. He discarded the rest. Not out of laziness, but because he understood that a badly labelled dataset is more dangerous than no dataset at all.

That is why I speak about this as a content producer, not an engineer. Content producers inject wrong labels into the ecosystem faster than anyone. An article with an attractive headline built on a mislabelled metric will be shared a few thousand times. The correction, if it comes, gets shared a few dozen.

"Why?"

Because a wrong label does not hurt. It only skews. And skew is invisible until a major decision is made on top of it.

Now to the part where I might be wrong.

There is a counter-argument worth weighing: labels may not matter that much. A good coach reads a player with his eyes, not with a statistics column. He sees how the player turns, how he positions himself, how he reacts after losing the ball. None of that lives in any label. If so, my entire concern about data labelling is a content producer's anxiety, not a football person's.

I also have to remind myself of another trap: using a European ruler on V.League. Being born in Germany and following European football since childhood makes it easy to assume a standardised data system is a prerequisite. But in a league where a club's analytics budget may be lower than half a season of a Bundesliga analyst's salary, hand-recording is not carelessness. It is a rational choice under given conditions. The right question is not "why don't they do it like Europe," but "with these resources, where should they audit labels first?"

And a third thing I might be wrong about: ambiguity in labels may actually be useful. A vaguely tagged player gets tried in more roles, and sometimes that is how a midfielder becomes a full-back, or a striker becomes a creator. Standardising too early can kill those possibilities.

"When the stands are empty, the match begins to speak its true voice."

I stand by my conclusion, but I narrow it. The problem is not whether labels exist. The problem is whether people admit their label is a decision.

My prediction for the rest of the season: at least one V.League club will publicly adjust how it records internal data after a defeat whose cause was misdiagnosed — most likely a match read as a physical failure, when the root lay in how their system labelled transition phases.

If I am right, you will see it within three rounds. If I am wrong, remember that I said clearly where I might be wrong — something a label line never does.

The crowd keeps cheering, and none of them knows whether they are cheering a label or a person.