Labeling Errors in Youth Football Scouting: When a Single Spreadsheet Row Convicts a Career

During a U17 screening at a training center in northern Vietnam in the summer of 2026, I struck a sixteen-year-old midfielder named Nguyen Duc Nam from the long-term watchlist. Three rows in the spreadsheet were enough for me to decide: a body-mass index below the standard threshold, a sprint speed under requirement, and an average distance covered of only about eighty-two percent of his age group. The spreadsheet did not calculate wrong. The error lay elsewhere: not one cell recorded that the boy had just returned from an ACL injury, and that his body was in the middle of a growth spurt that distorted every metric.

Three months later he made his first-team debut and recorded four assists in his first five matches. I do not retell this to flagellate myself for fun. I retell it because it repeats every season, at every academy, under a different name: a labeling error. And a label, once it sticks to a file, is harder to remove than people think.

The evaluation system runs like a labeling machine

Every football academy has to classify. A center with three hundred children and seven coaches cannot watch each one by eye. They need labels: potential, watch, cut. Labels allocate resources, decide who boards, who earns a training contract, who is sent back home. Without labels, the system collapses under overload.

The trouble appears when a label is born from a surface signal. In data processing, this is called a false positive from keyword collision: a word with several meanings, or a metric appearing in two different contexts, triggers a wrong classification, and the error then cascades down the entire chain. An article about cinema can be tagged as sport merely because it contains the word "lineup" or a place name that matches a club's name. Nobody rechecks the content, because the label has answered in place of the question.

In Vietnam the problem has its own color. Major centers such as Viettel, PVF and Hoang Anh Gia Lai have built scouting-data systems over more than a decade. But most of that data still sits in each coach's private spreadsheet, unstandardized, un-cross-checked. When a coach leaves, his label stays behind, while the context behind the label vanishes with him. His successor reads the file, sees the word "slow", and has no way of knowing in what circumstance that word was written.

After the feat of Vietnam's U23 side at the 2026 AFC U23 Championship in Changzhou, attention on youth football surged. The pressure to label surged with it, because every youth match now draws hundreds of thousands of viewers and commenters. The annual-season rhythm makes the disease worse. With a dense calendar, every round is a fresh chance for public opinion to attach a new label. A young striker who misfires once is called luckless. A defender who errs once is called mentally weak. New labels pile onto old ones, and within months nobody remembers where the first label came from.

Three layers of context beneath a single number

Statistics are the topsoil; I always dig three layers deeper.

The topsoil is the metric. That is what shows up in the spreadsheet: body-mass index, sprint speed, distance covered, goals scored. These metrics are real and should not be dismissed. But they are only the starting point of the excavation, not the conclusion.

Beneath the topsoil lies biomedical context. A sixteen-year-old may be mid-growth-spurt, when bones lengthen faster than muscle, temporarily degrading coordination and cutting sprint speed. The same player, once muscle catches up with bone, can recover every metric within six months. Injury leaves traces longer than people think: damaged ligaments change the running mechanism, and the body must relearn how to accelerate. Injury does not erase a talent's name; it merely sinks that talent into a sedimentary layer.

I once ignored this layer and paid for it. In 2026 I concluded that Nguyen Duc Nam lacked the physical foundation, ignoring that he had just returned from injury. I then had to add a full biomedical-context column to the data table, and from then on I no longer trusted dry numbers absolutely.

Deeper still is opponent quality and the denominator. A striker scoring eight goals in a lower division is not remotely equal to one scoring eight in the top flight. A midfielder with a high pass-completion rate may simply always pass backward. This is why I cling to per-ninety-minute output instead of total minutes. In 2026, reviewing an academy, I met an eighteen-year-old striker named Tran Van Cong whose per-ninety output was the highest in the setup, yet who cramped frequently and rarely played. By total minutes he was almost invisible. By output plus archived GPS data, he was the most efficient man there. I recommended a professional contract before the league resumed. The next season he scored six goals.

I learned to separate two kinds of signal: noise and truth. Noise is what appears in one match, one week, one run of form. Truth is what repeats across different contexts — home and away, strong and weak opponents, fresh and tired. A young player has a true signal only after you have watched him in at least three different contexts. Before that, every conclusion is a guess dressed up in metrics.

The same logic applies to risk data. In 2026 I reviewed a defender named Le Van Son during a transfer window. Three AFC Cup matches showed him winning twelve tackles, an impressive figure standing alone. But set beside three direct errors leading to goals under away pressure, the picture changed color. The tackle metric measures effort; it does not measure decision quality. I advised the club against a long-term deal. Two weeks later he suffered an injury and the contract was cancelled.

At a more macro level, effort metrics can also mislead. Distance covered and sprint counts are packaged as measures of character, but ineffective running still produces pretty numbers. A player covering twelve kilometers may simply be chasing a ball that has already passed him. A data map can point the wrong way if you do not read the terrain. So I always place effort metrics beside position metrics, and place both beside opponent quality.

In 2026 I tracked a young midfielder in a major league and found his distance covered fell eighteen percent after the seventy-fifth minute. I warned in my report that he would decline if pushed to extra time. The coaching staff did not rotate, and he left the tournament injured. I then realized I had been slow to adapt to the high-intensity trend, and began studying machine-learning models. But the larger lesson remained the old one: a single metric, however accurate, can still lead to a wrong conclusion without three layers of context.

Forgiveness does not exist inside a data table

The counterintuitive point is here: the problem of Vietnamese youth football is not a shortage of data, but a surplus of labels and a shortage of forgiveness.

In a justice system, a wrongfully convicted person can be exonerated after nineteen years. But even when the verdict is erased, public opinion remembers the accusation longer than the exoneration. The label clings to a person more stubbornly than the truth that they were innocent. Football works the same way. When a young player is branded a failure, the correction never draws the same readership as the original branding. The shock of an accusation spreads fast; the silence of an exoneration is reported by no one.

So I do not excavate stars; I excavate context. The data analyst's task is not to find the best child, but to find the child misjudged for lack of context. That is far less glamorous than hyping a prodigy. But it is more useful work.

There is an opposite temptation to guard against too: excessive humility. If every time I read numbers I said there was not enough data, I would never make a decision, and the club would have no need of me. Caution has value only when it leads to a concrete, testable action. Otherwise it is just a wordy way of dodging responsibility.

A hypothesis to test

The hypothesis I want to test over the next two seasons: if an academy adds a biomedical-context column and an opponent-quality column for every trainee, the number of resurrections — players once cut who return to the first team — will rise markedly. This is measurable. And if it holds, what needs fixing is not the scout's eye, but the structure of the data table they are using.