The Empty Report: A Lesson on Data Honesty in Football Analytics

When the skeleton looks better than the flesh

Last autumn I received a forty-page document from an analytics group I had worked with before. The cover page carried the name of a major fixture. The table of contents split cleanly into nine sections: tactics, club finance, form, league landscape, rules and governance, dressing room, risk profile, media narrative, and industry transmission. There were charts. There were tables. There was even a glossary at the back.

I opened the first page and read. No team names. No player names. No scoreline. Not a single xG figure, not one PPDA reading, not one transfer fee.

Every cell in every table carried the same line: insufficient information, cannot assess. Every conclusion pointed back at a data field that did not exist. The tactical matrix had no subject. The financial sheet had no balance sheet. The risk matrix had no risk. And at the end, in the recommendations section, the author — presumably an automated system — admitted plainly that the upstream input had been empty, and therefore no substantive football analysis had been produced.

I closed the laptop and opened it again. It was still empty.

What struck me was not the emptiness. What struck me was the structure. The report was so polished that a reader skimming it would never suspect. If I had only scanned the headings and nodded, I would have carried away the sense that someone had done serious work. But beneath that coat of paint there was no wall.

I once thought I had seen every kind of data error in this trade. I have seen data collected at the wrong moment. I have seen a data-entry clerk mistype a decimal point and turn a defensive metric into an attacking one. I have seen xG models inflated because they counted shots from absurd distances. But this error was different. This was an architectural fault: a complete template emitted before any content was loaded.

And the paradox is that this very emptiness taught me more than any full report ever has.

Context: the analytics industry and the template trap

I am forty-four years old. I was born in Vietnam, studied statistics, and moved to Shanghai to work for the Chinese football market. Twenty-eight years of watching football, and for the past five I have worked directly with coaching staffs and scouts. I am not a writer of emotion. I write with numbers.

But precisely because of that, I understand better than most that numbers do not generate themselves. Behind every metric sits a pipeline: an observer, a recording system, a data-entry operator, a checker, an interpreter. Each link can distort the number. And when a link stops running, the output is rarely the honest statement that there is no data — the output is usually an empty structure dressed in a complete-looking skin.

Modern football analytics has a built-in weakness: it loves templates. From xG models to PPDA frameworks to nine-part scouting reports like the one I received, everything is standardised. Standardisation is necessary, because it enables comparison and reuse. But standardisation carries a risk: once the mould exists, there is pressure to fill it, even with pieces that do not belong.

I once watched a small Chinese club prepare for a relegation battle using a three-hundred-page scouting file on a foreign striker. Every page had numbers. But when I traced the sources, I found that forty percent of the data had been copied from another player's report — only the name was changed. Nobody checked. Nobody questioned. The mould had been filled, and that was enough to make a decision.

That is what I call manufactured perfection. And it is more dangerous than scarcity.

Four categories of football data error

I divide football data errors into four categories, drawn from my own experience tracing and cross-checking. When all four appear together, they build a structure that is complete but hollow.

The first: the circular pointer. This is the most technically subtle fault. A data field is instructed to draw its value from another field — but that field does not exist. In the document I received, the related entities field was explicitly told to draw from the information points above. But the information points above were empty. So what were the related entities? Nobody could say. A closed loop. In football terms, it is equivalent to an xG model built from shot data, where the shot data is itself derived from the xG model. Formally elegant, substantively meaningless.

The second: the blank cell disguised as no risk. This is the most dangerous fault in live operations. When a club receives a risk report full of blank cells, it tends to read blank as clean. No injury data on a key player does not mean the player is fit. No financial fair-play finding does not mean the club is compliant. In the document I received, the entire compliance table said insufficient information. Had it been sent to a board, they would have read it as a clean bill of health. A fatal mistake.

My rule, after years of cross-checking, is this: absence of evidence must never be read as evidence of absence. I have to repeat this to young scouts constantly. Once, a colleague handed me a list of six wingers annotated as having no injury history. I asked to see the source. It turned out the database had not been updated for their league in two seasons. No injury history because there was no injury record — a gap, not a guarantee.

The third: skeleton perfection. A template generated before content is loaded. In analytics systems, this happens when an automated process runs the interface without waiting for input data. The result is a document with full headings, full sections, full tables, and empty cells. To a skimmer it is a piece of work. To a careful reader it is a confession. In football, this is the kind of report clubs routinely receive from subscription analytics vendors: glossy outside, thin inside. I once refused to sign my name to such a report and nearly lost the contract. But I kept the principle: better to say I do not know than to say I analysed something I never analysed.

The fourth: circular reference data. This is a fault at the time layer and the source layer. A report with no publication date, no source, no timestamp for when the data was collected. In the document I received, both the time-sensitivity field and the source-quality field were left unassessed. That means even the season the story belonged to was unknown. In football, that is equivalent to being handed a PPDA reading without knowing which match, which league, which season. That number could be the signature of a ferocious pressing side, or it could be one match played in heavy rain. Without a timestamp, there is no conclusion.

Anatomy of nine empty dimensions

When a nine-part analysis is empty, each part is empty in its own way. I reopened the document and walked through it.

Dimension one, tactics and technique: no formation, no system, no style. The tactical assessment table has four rows — sophistication, execution, personnel fit, key data. All four read insufficient information. This is the most direct fault: no subject, no analysis.

Dimension two, club finance and the transfer market: no broadcast revenue, no commercial revenue, no wage bill, no net debt. In modern football this is the most important dimension of all. A club can win on the pitch and go bankrupt in the accounting room. But with no club name, every number is a zero.

Dimension three, results and the public-opinion cycle: no standings, no form sequence, no pressure. I do not know whether this is a title race, a European chase, or a relegation fight. For an analyst, that is total blindness to context.

Dimension four, league landscape and team positioning: the tier diagram from title contenders to the relegation zone is empty. No league named. No comparison group. No talent-flow signal.

Dimension five, rules and governance: the compliance checklist covering financial fair play, transfer registration, discipline, and competition eligibility is blank. The applicable rule system cannot be identified.

Dimension six, management and dressing room: no owner, no sporting director, no coach. No quotes, no social-media signals. The dressing room is a black box.

Dimension seven, risk profile: the matrix has six categories — sporting, financial, personnel, rules, public opinion, systemic — and all six are empty. But this is the most dangerous dimension to leave blank, because readers mistake blank for no risk.

Dimension eight, media narrative and expectation: no storyline, no protagonist, no event. The heat phase of the news cycle cannot be located.

Dimension nine, industry transmission: the diagram from academy to club to derivative markets is empty. No upstream node, no downstream node.

What stands out is this: all nine dimensions are empty, yet the template remains complete. That is the lesson. A template will never protect itself against emptiness. Only a careful reader can.

Three field stories from the trade

The three stories below show that these faults are not theoretical.

In 2026, when Hulk moved from Zenit to Shanghai SIPG for a fee the media reported as 55 million euros, the entire Chinese market erupted. I was thirty-five and working as a data analyst. I built a cumulative xG model for Hulk across three seasons in Russia, adjusted for league quality, and compared it with the media expectation. The result: his actual finishing output was around 0.28 goals per match, nearly forty percent below the figure the press had pushed out. I wrote the report and published it. Fans attacked me hard. But three scouts from other clubs contacted me for the detailed version. I learned something: accurate numbers will find the people who need them.

But that story also taught me the opposite lesson. If I had not had the underlying shot data that day, I would have faced a choice between two extremes: either say there is insufficient information and be seen as weak, or invent a number to protect my reputation. Many people in this trade choose the second. And that is exactly the manufactured-perfection fault I described above.

On 27 June 2026, while commentating live for a television station during Germany's match against South Korea at the World Cup, I issued a warning based on Germany's PPDA in their game against Sweden. The figure was only 7.8 — about thirty percent below their group-stage average. I said that if Germany kept pressing lazily, they would lose. The lead commentator laughed at me. Viewers called in to insult me. Then Kim Young-gwon and Son Heung-min scored, the scoreline became 0-2, and I became a viral phenomenon.

But the story behind it that few people know: the night before the match, I had to decide whether to make that call at all. My dataset was incomplete — I was missing PPDA for the Mexico game, and some of the shape-distance metrics were corrupted. Had I obeyed the template, I would not have dared to state a conclusion. But I had three advanced metrics that were strong and consistent: PPDA, opponent xG from counter-attacks, and Germany's midfield line distance in the second half. Three data points were enough to form a signal. I chose to speak, I flagged the data limitations clearly, and I let the audience judge.

From then on I set a rule for myself: every match analysis must contain at least three advanced metrics as evidence, and must state the data limitations. I never write by impression alone without numbers. And I never invent a number just to fill a template.

In 2026, when the pandemic forced leagues worldwide to play in empty stadiums, I collected Premier League data from 2026 to 2026 and compared it with the post-lockdown sequence. The result: home win rate fell from 46.2 percent to 38.4 percent, while average goals per match rose by 0.6. I wrote a forty-page report and sent it to a club fighting relegation. They hired me as a set-piece analytics consultant — work that does not depend on crowds. I gave up my media-expert role to work directly with the coaching staff.

But the point I want to stress is not the achievement. The point is that during that period, a great number of football analytics reports were produced without crowd data — a core variable. Instead of stating that crowd data was absent, many reports kept the old template and filled it with conclusions built on assumptions. That is the circular-pointer fault at industry scale.

The line I use with students is: the stadium was empty, but the data never lost its crowd. Crowd data does not sit in the stands. It sits in how players move when there is no roar driving them. That is the purest signal of a team's true nature. The catch is that to read it, you must accept that your data is imperfect.

Lessons for Vietnamese football

I was born in Vietnam and I have always followed the V-League with an analyst's eye. In recent years Vietnamese football has made real progress on data. Clubs are starting to hire analysts, matches are tracked with more granular metrics, and youth academies are adopting data-driven methods.

But I also see signs of manufactured perfection. Some internal scouting reports are built on international templates but lack source data. Some models are imported wholesale from Europe without adjustment for the specifics of the V-League — a different match tempo, different pitch conditions, different player psychology.

I once reviewed a report on a talented young player from a northern academy. It ran twenty pages, included a radar chart, and compared him with European players of the same age. But when I traced the data, I found the comparison sample contained only four players, and two of them played in the third tier. A sample too small to support any conclusion. If a club used that report to price him and make a transfer decision, they could buy the wrong player at three times his true value.

This is why I believe Vietnamese clubs, before buying more data, should build an internal validation process. A process that asks: where did this data come from? Is the sample large enough? Has it been adjusted for the league? And if something is missing, are we willing to say there is insufficient information rather than invent a conclusion?

The answer to that last question determines the quality of the whole game. A match lasts ninety minutes, but its story lasts longer than a season. And that story can only be told correctly if the teller is honest about what he knows and what he does not.

The contrarian angle: scarcity is worth more than fake abundance

There is a widespread belief in football analytics: the more data, the better. Clubs spend millions buying data warehouses, hiring analysis teams, building dashboards. But twenty-eight years of experience tell me something different. The most expensive thing is not data. The most expensive thing is honesty about not having data.

I once saw an xG model presented at an analytics conference in Shanghai. It had a high coefficient of determination, elegant charts, and was advertised as predicting match outcomes with seventy percent accuracy. But when I asked about the training data, the presenter admitted they had used data from a single league over two seasons. A beautiful model, but it learned from a small sample, and it would collapse the moment it met another league. That is manufactured perfection dressed in the robes of science.

By contrast, some of the empty reports I receive are, precisely because of their emptiness, strangely useful. When a template admits it is empty, that is a signal. It tells me the report-production process has a fault. It tells me the person or system that generated it did not fabricate content. It tells me that if I re-run the pipeline with proper input data, I can get a substantive result.

Compare that with what I call the beautiful-but-hollow report — documents filled with unsourced numbers. These are far more dangerous, because they create a false sense of security. A club that reads such a report will make transfer decisions, build tactics, and bet on numbers that are not real. When it fails, it will not know why, because the report still looks beautiful.

This is why I keep repeating to younger colleagues: do not trust a number until it has told its story from the beginning. Before using any metric, I trace its origin. Who collected it? When? By what method? Has it been adjusted? If I cannot answer, I set it aside. I do not delete it, but I do not build conclusions on it.

Here is another paradox: many people assume this caution makes me slow and indecisive. In reality, it is precisely what allows me to make counter-intuitive calls that turn out right. When I know exactly how limited my data is, I know exactly which limits can be crossed and which cannot. I know when three metrics are enough to conclude, and when twenty metrics are still not enough.

In football, every number has a story behind it. A low PPDA could signal a lazy pressing side, or a side deliberately ceding territory to counter-attack. A negative xG conversion rate could signal an unlucky striker, or a system that does not create enough quality chances. If you read the number without understanding the story, you will use it wrongly. And if you invent the story to fill the number, you destroy both.

In the empty report I received last autumn, there was one line I want to quote as a reminder: the blankness of the table does not mean no risk was found. It means there was no information to search. That is the distinction every football analyst must burn into their memory. Because football, more than any other sport, lives on emotion. And emotion always wants an answer. When there is no answer, emotion will invent one. The analyst's job is to stop that invention from happening.

I remember a time when a coach asked me: is this player fine? He wanted a number. I said: I do not have enough data to answer with certainty. But I have three metrics saying he is declining, and one saying he is merely in an adjustment cycle. My view is that we should watch two more matches. He was not satisfied. Two matches later, the player was injured. If I had invented a number and said he was fine, the coach would have played him, and the consequences would have been heavier.

A forward-looking thought

The rise of artificial intelligence in recent years makes this problem more urgent. When large language models are used to write football analytics reports, the manufactured-perfection risk multiplies. A model can produce a thirty-page report with full headings, full tables, full conclusions, without any real data at all. It will not confess the way the document I received did. It will fill every cell with something that sounds plausible.

That means football analytics needs a new fence. Not a fence to block more data, but a fence to test the honesty of data. A validation gate at the input: if core information fields are empty, the system must stop, must raise an error, must refuse to generate conclusions. No template should be allowed to complete when the input is empty.

As a data analyst, I believe the future of this trade lies not in more complex models, but in more honest ones. When probability collapses, what remains is the nature of the match. And the nature of the match is never written by invented numbers.

We are midway through the season, when every passing week spawns hundreds of analytics reports. Most will look beautiful in form. Some will be correct in content. But only a very few will be honest in both. Those are the reports worth reading. Those are the reports I, as someone who writes with data, want to leave behind.

As I keep telling those entering the trade: data never gets tired; only the person reading it does. But data will never protect itself against fabrication either. Only people — or processes — can do that. And when you keep that honesty, you are not just analysing football better. You are protecting your own profession.

The question I would put to every football analyst this week: what percentage of the numbers in your most recent report can be traced back to their origin? If the answer is below fifty percent, then perhaps you do not need more data. You need fewer numbers that are not real.