Knowing You Don't Have Enough Data: The Hardest Lesson VAR, the Transfer Market, and Esports Are Relearning

Minute 78, Thong Nhat Stadium. A visiting defender slid in for the ball inside the penalty area, the home striker went down, and the referee pointed to the spot within a single breath. In the VAR room nearly a hundred meters away, the video officials put their hands to their headsets. Four screen angles divided a moment less than two seconds long. The footage slowed, each tenth of a second running backward. Then no one pressed the button. The referee kept his decision, the match continued, and the stands split into two halves of fury.

I sat high in the stands, a small notebook in hand, recording the number: the VAR team's response latency that night was 6.4 seconds, close to FIFA's recommended threshold. But the detail worth recording was not the number. It was that nobody in that closed room dared to say one simple sentence: "We don't have enough camera angles to conclude."

Nine years in the job, from a VAR assistant role in Incheon to analysis sessions for referee committees, I learned something no school teaches: the hardest skill of a person holding the scales of judgment is not catching a mistake, but knowing when there is not enough evidence to call one. That holds for a VAR room, for a transfer scouting department, and for an esports analytics team reading data on an opponent.

Vietnamese football has moved past the phase of memorizing the rules; it is entering a harder phase: learning to refuse conclusions when the data is not enough.

When VAR first arrived in major Asian matches, most controversies did not revolve around referees blowing wrong calls. They revolved around referees blowing correct calls and failing to explain why. Fans see a player fall and decide within 0.3 seconds. The VAR team sees four camera angles and takes 40 seconds to reach no conclusion at all. Those two speeds of perception cannot be reconciled by a whistle.

I once thought speed was the answer. If the VAR team responded faster, the stands would calm down. In 2026, when I was still a VAR assistant in the K League, I believed that. I believed that all I needed was to optimize the process, cut the extra steps, and every controversy would dissolve. That mistake cost me three sleepless nights and a naive faith in numbers.

That night, at round 29 of the K League, in minute 67, Lee Dong-gook scored for Jeonbuk Hyundai Motors against FC Seoul. I spotted him offside by 0.3 meters. But I was too absorbed in reviewing the rear camera angle and sent the alert signal 14 seconds late, double the seven-second standard. The main referee had already let play continue. The goal stood. The managing director scolded me in front of the whole editorial room. I did not resent him. I resented my own fourteenth second.

For a week afterward, I sat breaking the footage down like a man picking a lock. And I found something that made me shiver: most controversial situations are not situations lacking data. They are situations with data, but data placed in the wrong spot. I did not misjudge the offside. I cut the wrong frame. I used the rear angle to find a player's position when the rear angle could never show me that.

The most common VAR error is not misreading a number, but using a correct number for a question that number was never designed to answer.

This is the crack I call the first crack. And it repeats everywhere, only with a different name.

In 2026, sent to Russia as a VAR analysis assistant for a Korean broadcaster during the World Cup, I confronted the second crack, this time one of definition. IFAB was trialing a new interpretation of handball inside the box. From the Spain-Iran group match onward, I began collecting data on every handball situation. By the end of the tournament I had 27 situations in hand. I checked them against the written law and found that only 31 percent were handled consistently under the same interpretation. Nearly seven in ten situations did not follow the same standard.

I wrote a 40-page report and sent it to the editorial desk. They ran a tiny chart in the corner of a page. Its headline read: "There is a difference in handling." That was all. No sentence from the report appeared. Frustrated, I spent one evening setting up a personal blog, published all the data with the raw spreadsheet untouched, and two days later the post spread everywhere. Fifty thousand reads, from referees to sports lawyers, even to passionate fans who argue on forums.

The first lesson I learned was not that data has power. It was that a truncated version of data does more harm than an honest version that gets ignored.

But that was only half the story. The other half lay in what I did not write.

What I did not write was this: 27 situations is far too small a sample to judge a definition still under trial. Thirty-one percent does not prove IFAB wrong. It proves IFAB issued a definition that was not clear enough for a tournament with only 27 samples. But once the blog spread, most readers skipped the phrase "not clear enough" and read it as "IFAB is wrong." A vague definition became a firm accusation, through a single share.

Every VAR error is a crack in the mirror that reflects the rules. And every time the community reads a crack as a crime of the institution, the mirror cracks a little more, not because the law is weak, but because faith in the law is being misread.

In 2026 I moved into transfer data. The consultancy where I worked built a player-evaluation model based on VAR data. The idea was seductive: if VAR measures every error, every play, every position, then we can use the same data to measure a defender's value. In 2026, my model produced a result that made me confident. Kim Min-jae, then at Napoli in Serie A, committed 0.73 fouls per match. On our scale, that was "high card risk." I advised the firm not to place him on its recommendation list.

Napoli signed him anyway. And he became a pillar who helped the club win the 2026 Serie A title.

I sat still for a long time with that result. I did not sit still because I was wrong. Anyone can be wrong about a player. I sat still because the 0.73 I read out was far too confident. My model counted fouls, but could not count a teammate's cover, could not count how differently Italian referees interpret the law from Korean referees, could not count that a defender in a three-man system can commit more fouls yet be punished less. I overlooked every variable outside the frame.

At the end of that year I wrote a ten-page self-review and took the model down. Not because the model was useless. Because I had used it to answer a question it was never designed to answer.

The transfer market has an unwritten law: it rewards those who dare to conclude quickly and punishes very late those who conclude hastily. A 0.73 goes into a report, passes through three levels of management, reaches a closed meeting, and becomes a decision. If my model is right, I am a genius. If my model is wrong, no one is punished, because no one bears personal responsibility for a probability.

This is why I am allergic to what people call the "young-player price bubble." A 19-year-old who has not played 50 top-flight matches gets valued at 100 million euros. That number is not born from one person's mistake; it is born from a long chain of people reading data without asking whether the data is enough. Everyone holds one fragment; no one holds the picture. And when the bubble bursts, the question always asked is "why is this player so bad," rather than "why did we price a boy with a number we knew rested on nothing."

In March 2026, global football paused. The broadcaster I worked for cut my contract to cut its budget. I did not rush to find a new job. I withdrew into research. For six months I spent my time analyzing 1,247 VAR decisions from five major European leagues, all played in empty stadiums. I wanted to know: with no crowd noise, do referees decide differently?

The result stunned me. Without crowds, the time referees spent consulting VAR fell 22 percent, but the rate of upholding the original decision rose 15 percent. In other words: when no one was jeering, referees reviewed less, yet trusted their own calls more.

I wrote a 60-page report and posted it on an academic forum. A director at the Asian Football Confederation reached out and invited me to work as a data analysis expert for a refereeing committee. That was the turning point of my career. But what stayed with me was not the invitation. It was the paradox I had just measured.

The noise of the stadium is not written into the law, yet it carries legal weight. It forces referees to slow down, to review, to question themselves. Remove it, and you think you have removed a pressure, but in fact you have removed a self-correction mechanism. The stands do not make referees more correct. But they make referees more careful. And that carefulness is itself a form of fairness the law cannot spell out.

In mid-2026, when I lost my contract, I learned the most important thing of my career: how to present a hypothesis, a method, and the limits of your own research. But I also confess I fell into another trap. I wrote phrases like "T-test verification" without explaining them. I trusted the rigor of my method so much that I forgot my readers had no obligation to read a scientific paper to understand a football match.

That was the third crack.

The natural position is the term I use most when training young referees. When judging a play, we do not ask whether a player is morally right or wrong. We ask: in that situation, at that speed, along that trajectory, was the player's body in a natural position? Is an arm extended to keep balance natural, or is it reaching out deliberately? This is a question of movement anatomy, not of intent. But Korean, Vietnamese, and international bodies explain this concept in three different ways, and the difference is no one's error. It is the difference between cultures reading the law.

I once watched a match in which a Vietnamese defender was penalized for a handball while turning his body. In a European league, the same motion might not be called. In Korea, it might be called but only warned. Three decisions, three mirrors, three different faiths in the same wording. Which one is "natural"? No one can answer. And because no one can answer, every controversy stops at emotion and never reaches the law.

This is where I grow suspicious of institutions. Not suspicion that they are corrupt. I suspect they hold a structural interest in keeping the concept vague. A clear rule is easy to verify. A vague rule can be applied in many cases without ever being wrong. And when it is never wrong, no one has to be accountable.

I set up a personal blog to publish all my data without asking anyone's permission. A slightly reckless decision, and I still wonder whether I would do it again.

But here is the hardest part, and the part I learned latest.

In 2026 I began extending my analysis into esports. In that world, a player's career span is far shorter than a footballer's. An esports pro can peak at 19 and retire at 24. Yet esports' youth-development and post-retirement support systems barely exist. No training academies, no pension funds, no career-transition network. An eighteen-year-old can earn a decade's income in two seasons, then be broke three years later.

What I observe in esports are the very mistakes I made, compressed into a far shorter time. A team analyzes an opponent's last three matches, spots a pattern, and treats it as truth. Three matches is far too small a sample. Three matches cannot separate signal from noise. But the pressure of competition gives them no time to wait for a bigger sample.

I once asked myself whether esports is the worst environment for data analysis. Then I realized: no. Esports is merely where every mistake is exposed faster. Football has 90 minutes and a week between matches. Esports has 30 minutes and 24 hours. In football, a wrong data decision can take a season to surface. In esports, it surfaces in the next game.

And here is the most surprising thing I learned comparing the two worlds: the problem is not that esports has less data than football. The problem is that esports has more data but less time to digest it. An esports match generates hundreds of thousands of data points per minute. A football match generates thousands. But esports' feedback loop is so short that humans cannot turn data into understanding in time. They are forced to guess. And when forced to guess, they use intuition and call it analysis.

VAR was born from the fear of making mistakes, but it breeds the fear of a truth that comes too late. In esports, people run into the same fear under a different name: the fear that if they do not win the next game, all analysis becomes meaningless. And when that fear leads, people no longer analyze. They only look for an excuse.

This brings me to a view contrary to common intuition.

People often think the problem of sports analysis is a lack of data. The real problem is too much data and too little discipline to refuse it. Anyone can extract a number. Very few dare to say that number cannot answer the question being asked. And the danger is that a wrong number, beautifully presented, is more persuasive than a correct silence. Silence does not make a slide. It does not make a headline. It does not make a thick report a boss can proudly present.

I have been in meetings where people argued for an hour about a wrong VAR decision, and no one spent a minute asking: can this camera angle show us this? The room was full of data. And the room was still short of information. That is a distinction not everyone can make.

Data is what we have. Information is what we can answer our question with. The two are not the same.

A wrong decision does not destroy a match; the silence after it destroys trust. A referee who errs, admits it, and explains, will anger people who still accept it. A referee who errs, while the institution defends him with a vague statement, will make people lose faith in the whole system. The difference is not in the error. It is in how the institution faces the error.

I have seen this in both football and esports. A decision that a crowd deems unjust will cool down within two weeks if the institution publishes its data. It will never cool down if the institution stays silent and waits for people to forget. The crowd's selective memory outlasts the institution's patience. And when an institution loses a contest of memory, it loses more than a match.

This is where I must be careful with myself. Because I was born in Vietnam and work in Korea, I am easily pulled to one side. A Korean referee makes a decision favoring a Vietnamese team, and part of me wants to believe it is fair. A Vietnamese referee makes a decision favoring a Korean team, and another part of me wants to hunt for the error. Both tendencies are wrong. A good referee must stand outside every flag, including the flags hanging in his own heart.

I once sat in a restaurant in Incheon, listening to an old man call the referee a thief, and I stayed silent. I stayed silent not because I agreed. I stayed silent because I knew that whatever I said would be read through the lens of where I come from. That feeling is strange. You are an analyst, but you are read as a fan. And you must choose between speaking up and losing your audience, or staying silent and losing yourself.

This is why I always add a "data limitations" section at the end of every piece. Not to shield myself from criticism. But to remind both myself and my readers that a conclusion holds only within the range of the data that produced it.

Intellectual humility is not weakness. It is a form of discipline. Everyone wants to say "clearly." Very few dare to say "I don't have enough data to know." But it is the second sentence that separates the analyst from the purveyor of words.

Back to the VAR room at Thong Nhat that night. I believe the people in that closed room were right not to press the button. They lacked enough camera angles. They lacked enough grounds to overturn a clear decision. On the law, they were right.

But I also believe they made another mistake. They did not explain. They let the stands fill in the blanks, write a story the data never supported. In modern sport, silence is no longer neutral. Silence is a statement.

If I have one proposal to improve refereeing in the years ahead, it is not more cameras, not semi-automated offside technology. My proposal is far cheaper. It is this: after every controversial decision, the institution issues a short, public explanation of why it was upheld or overturned. Not to defend itself. But to teach fans to read the law the way referees read it.

This will disappoint many. Because it admits that there are times we cannot know for sure. But honesty about the limits of perception does not weaken football. It matures it.

I still remember a veteran referee telling me after a meeting that in 20 years of officiating, the hardest thing he learned was not reading the law, but recognizing when he was reading the law through his own ego rather than the original text. I think of that line every time I sit beside a screen full of data.

When you have data, it is easy to believe you have the truth. But data never speaks on its own. It only speaks when we place it inside a frame of questions. And if the frame is wrong, more data takes us further away.

This is why I took down the Kim Min-jae model, even though it was only a model. Not because it calculated wrong. Because it gave me an answer to a question I should not have asked.

In esports, this repeats every day. A team reviews three lost games, finds a pattern, and believes it is the key. But three games is not a trend. Three games is a phase. And a phase always changes.

I wonder how many failed transfers in world football originate from a well-designed model that answered the wrong question. I suspect that number is far larger than the failures caused by players genuinely playing badly.

There is one thing I still have not managed, and I say it here to remind myself. I still write at length. I still prefer to narrate data rather than condense it. I am still caught between wanting to be accurate and wanting to be read. I do not think I will resolve this tension overnight.

But there is one thing I am sure of. In every field I have touched — VAR in football, transfer-scouting models, esports data analysis — the final winner is not the one with the most data. It is the one most honest about what their data lacks.

The mirror of rules will have its cracks. It will shatter under more camera angles than anyone can fix in time. But a mirror that knows it cracks still reflects light better than a mirror patched with the naive faith that it never cracks.

And if we keep demanding absolute certainty from a profession built on reading situations within tenths of a second, then we are not seeking justice on the pitch. We are seeking an excuse to stop arguing.

That excuse does not exist. And perhaps that is the only thing in this piece I can say without adding a "data limitations" section.