The Empty Data Frame: The Cost of a Football Conclusion Without Evidence

In the control room of a sports broadcaster, the third screen from the left is always mine. That night it was a qualifier, the clock on the wall was ticking past the 70th minute, and the data board I had requested from the tracking system returned an empty frame. No xG. No PPDA. No count of presses within five seconds of losing the ball. Just a grey error line, the kind every data consultant sees at least once in their career. The presenter turned to me, mic still open, and asked: "Anything for the second half?"

In that moment I faced a choice that modern football analysis confronts every day, only on a much larger scale. I could produce a plausible-sounding judgment based on what my eyes had seen. Or I could tell the truth that the system was empty. I chose the second. For seven minutes afterward I sat silent and let the presenter handle his own emotional commentary — his job, not mine.

That incident was minor. But it is a miniature of a problem eating away at the entire football analysis industry: when data does not exist, people still produce conclusions, and those conclusions carry the full authority of a field that lives on the word "data".

Today I want to talk about that empty frame. Where it appears, why it is more dangerous than an ordinary error, and why in football, an analysis built on empty data has a destructive power greater than a merely wrong opinion.

In 2026 I joined the sports desk of Belgrade Television in my early twenties. Back then "data" meant a notebook and a pencil. I counted a central midfielder's passes, recorded the position of each shot, labelled every foul. Forty years later I sit in front of twelve parallel data streams, each with its own sampling rate, its own vendor, and each capable of dying at any moment. The tools changed completely. But the 2026 notebook and the empty data frame of the other night share one lesson: if there is nothing on the page, everything written afterward is fabrication.

That is why I am addressing something that sounds purely technical, when it is really a story about professional ethics. Vietnamese football is entering a phase where every V.League match is captured as thousands of data points. But quantity does not equal quality. And the quality of a data board starts with the simplest question of all: is anything actually there?

I once watched an analytics assistant present a report on a player built on "pressing metrics" when the club's tracking system had lost a camera from the first half. He did not know. He pulled numbers from an online aggregator, unverified, uncross-checked. The report ran nine pages, printed in colour, bound. The head coach read it and made a decision. That was when I understood the problem is not missing data. The problem is fake data presented in the same format as real data.

A wrong number can be fixed. A number with no origin cannot, because nobody knows where it began.

Data never lies, but the people who read it do. I have written that line many times, but I have never spelled out its most dangerous version: the liar does not need to invent numbers. He only needs to present a number with no provenance and let the authority of the word "data" do the rest.

Look at how a scouting report is built in modern football. We start with a few seemingly objective indicators — goals, assists, pass completion. These are easy to find, easy to copy, and almost always available on aggregator sites. The problem is that an uncontextualised indicator carries almost no information. A centre-back's 92% pass completion at a possession-dominant team says nothing about his defensive ability. Goals say nothing about the quality of chances he creates.

When an analyst has no data provenance, he fills the gap with assumption. And assumption, written in the language of statistics, becomes collective belief. This is the mechanism I call data dogmatism — when a number's authority is inherited from an entire industry's authority rather than from its own evidence.

World Cup 2026 taught us that emotion is the hardest noise to filter. But it taught another, less-cited lesson. Throughout that tournament I sat in a broadcaster's data room and watched the same phenomenon repeat every round. When a big team was eliminated, the industry immediately produced a wave of "decoding the failure" reports. Most were written within 24 hours, based on readily available aggregate metrics, with no positional data, no speed tracking, no defensive structure analysis. And they all concluded in the same direction: inefficiency in finishing.

It is a correct conclusion. It is also meaningless, because every losing team is inefficient at finishing. It cannot distinguish a team eliminated by a faulty pressing structure from one eliminated by an outstanding opposing goalkeeper. It has no predictive value. And it was born from an empty dataset, because the writer had no access to anything that could actually differentiate the two situations.

In football analysis, a conclusion without data is not an incomplete conclusion. It is a false conclusion presented as complete, and that is the most dangerous kind.

I spent three weeks after World Cup 2026 reviewing all 64 matches on tape, cross-checking data against reality, and producing a 200-page document on fatigue-index forecasting. That work taught me a principle I keep to this day: every number is a confession, if we are patient enough to listen — but only when that number speaks from a traceable source. A number with no source confesses nothing. It just stands there, silent, waiting to be assigned a meaning.

Let me get concrete with an example I lived through. In the 2026 V.League season, when I was a data consultant for a club in Saigon, I built a system tracking twelve movement metrics per player. High-intensity distance, presses within five seconds of losing the ball, and the rate of passes into the final third. In the round-18 match against Hanoi FC, my system recorded young midfielder Nguyen Trong Huy running only 8.2 km in 90 minutes, 15% below the team average, with only four presses within five seconds against a positional average of eleven. I recommended substituting him at the 60th minute.

The coaching staff ignored it. The team lost 1-3. After the match I presented a 14-page analysis with charts, cross-checks, and clear sourcing for every metric. From then on the head coach began following my data-driven adjustments. The club finished the season fifth, four places above the pre-season projection.

I tell this story not to boast. I tell it to make a point: the power of that report was not in its twelve metrics. It was in the fact that every metric was traceable to a specific passage of play in a specific match at a specific minute. Had I simply written "Nguyen Trong Huy ran little", the report would have been dismissed like any other opinion. What gives data weight is not the data. What gives data weight is traceability.

And here is what I want those doing football analysis in Vietnam to remember: a data board with no provenance is not data. It is literature, and bad literature at that.

Here I must address a topic the industry finds uncomfortable: AI and large language models flooding into football data rooms. I watch matches, and I have no aversion to technology. On the contrary, I believe machine-learning models can handle problems humans cannot — optimising fixtures for a team in three competitions, or analysing thousands of hours of video to find movement patterns the eye misses.

But there is a problem. AI generates fluent text from empty data. That is its nature. If you ask a model to analyse a match for which it has no data, it will not return an empty frame. It will return a smooth, structured paragraph full of technical jargon and completely devoid of evidence. A general reader cannot tell that paragraph from a real analysis. That is the danger.

I tested this over months. I gave a model an empty match summary — no team names, no players, no metrics. The output was a complete tactical report with sections on xG, PPDA, pressing structures, and transfer recommendations. It was wrong on every count. But if I had not told you the input was empty, you could not have detected it. A professional fabricator does not need to lie. He only needs to be fluent.

This is why I always close my analytical recommendations with an open question about the reliability of the statistics. Not to seem humble. To remind myself, and the reader, that being 62 has not slowed me down; it has taught me which data is worth waiting for.

Euro 2026 gave me another example, this time directly relevant to Vietnamese football. In 2026, studying the tournament's effect on Southeast Asian players' workload, I found that Vietnam's national team had six players who had played over 2,800 club minutes before entering World Cup qualifying. I sent a recommendation to the federation to manage Quang Hai's load against the UAE in the group stage. All of it was ignored. Quang Hai suffered an ankle injury in the 23rd minute against the UAE; the team lost 0-1 and lost its path deeper into the campaign.

I then collected data on 40 Southeast Asian players at the Euros and Tokyo Olympics myself. The result: 57.5% of them declined an average of 18% in form within two months after the tournament. The report was later used by a German researcher in an article on post-tournament syndrome.

The Euro 2026 injuries were not a curse; they were a delayed report. But I want to push one step further. Had my 2026 recommendation been written without specific minutes data, without comparison to prior tournaments, without clear sourcing — would it have carried any weight? The staff would have dismissed it, and they would have been right to, because advice without evidence cannot be distinguished from a rumour.

The scariest thing in football analysis is not a wrong conclusion. It is a correct conclusion presented without evidence, because it discredits every other correct conclusion.

Now I want to address the other side of this story, the part data-first analysts like me are reluctant to admit. My whole argument so far has a weakness. If I demand that every conclusion have provenance, I am assuming provenance always exists, is always reliable, and is always read correctly. None of those three is automatically true.

Look at the leading data providers themselves. Opta, StatsBomb and similar names have error margins. They have calibration processes, cross-checks, confidence thresholds. But even positional data is not absolute truth. A pass near the touchline may be logged as successful by system A and unsuccessful by system B, depending on how "the final third" is defined, where cameras are placed, how the algorithm detects the receiver. Cross-check two data sources on the same match and you will find discrepancies in roughly 3% to 5% of passes. That sounds small, but it is enough to flip the conclusion about a player whose pass completion sits near the average threshold.

And even with perfect data, the reader can still be wrong. Correlation is not causation. Every analyst knows that line; few truly apply it. A team with more passes usually wins more. Does passing more cause winning? No. More likely the reverse: strong teams hold the ball more. Read data without distinguishing causal direction and you will build a tactic on effects rather than causes. Many poor transfer decisions in world football stem from exactly this error.

The transfer market is the only place where people pay for hope, not output. And once data becomes a pricing tool, we get a system where a misread metric can make a club pay an extra million dollars for a player whose true value is lower. I have seen it happen. I have seen a young player overvalued on the back of goal numbers in a league whose defending is two grades below the destination league. The buyer read the goal metric and ignored the context. They bought a number. They did not buy a player.

Data is a mirror; a fool looks into it and sees himself, a wise man sees the team. I write that for the reader, but also for myself. Because I have been the fool. In 2026 I believed in a metric set I had built and thought it could predict muscle injuries. It correctly predicted about 60% of cases. I thought that was a success until a colleague pointed out that if I predicted every player was at injury risk, the hit rate would be 100% and the information value zero. I had confused being right with being useful.

Defence is not cowardice; it is the expression of probabilistic intelligence. I still hold that line. But I must add a clause: probabilistic intelligence only has value when the probability is computed from real data. A probability model built on empty data is not a model. It is a game of chance dressed in mathematics.

Here is where I say something the industry finds uncomfortable. Contrarian thinking against the majority is the duty of the person holding the data. But contrarianism without data is arrogance in professional clothing. I have seen people use scepticism as an identity, rejecting every prevailing conclusion without offering any evidence in its place. That is not contrarianism. It is a form of intellectual bullying. And I want to distance myself from it decisively.

If the majority is right this time, am I willing to admit it? That is the self-check I write for myself before offering any contrarian view. If I am unwilling to write out the answer, I have no right to offer the view. Forty years of watching the industry have taught me that the longest-surviving analysts are not those who are right most often. They are those who correct themselves fastest.

And to correct yourself, you need something this industry increasingly lacks: a data source traceable to the very end. Not a number. A trail. A path from conclusion back to the passage of play, to the minute, to the match, to the recorder, to the device, to the timestamp. With that trail you can verify, refute, and correct. Without it, you are reading a story, not an analysis.

What I want to leave here is not a technical tip. It is a change in how we read. Next time you read a football analysis, do one simple thing: find the source of two numbers. Not five. Two. If you cannot find them, put the article down. You do not need to prove it wrong. You only need to notice it has nothing to prove.

Vietnamese football analytics is at a fork. Clubs are investing in tracking systems, broadcasters in data graphics, and young players are growing up in an environment where professional metrics become part of a career. If we build this culture on sourced data, we will have a decade of progress. If we build it on fluent numbers with no provenance, we will have a decade of wrong decisions presented in the language of precision. The difference between the two scenarios is not technology. It is the habit of asking questions.

I still sit at the third screen from the left in the control room. The data frame that night is still empty, and it will be empty many more times. The only thing that changed is how I respond to it. I am no longer ashamed to say "I have no data". In an industry where everyone wants an immediate answer, daring to say you do not know is the most accurate form of data. It is the only number that needs no source to be honest.

As for the signal for the next round: watch which analyses cite no sources over the next two months. Not to catch them out. To see whether we, as a football culture learning to read numbers, are ready to tell evidence from impression. When the answer is yes, we will no longer need to believe in numbers. We will begin to understand them.