Football Data Failure: When 'No Information' Is Read as 'No Problem'

It was 2:47 a.m. in London. I was sitting in front of a screen with a spreadsheet already open, a cold cup of coffee, and a deadline knocking at the door. I was about to fire off a take about a Premier League match I had only watched one half of. My usual data source returned a page so clean it felt cold: every metric was zero. No warning. No connection error. Just wide-open empty cells, columns of nothing, and a flat line looking like a heart that had just stopped.

After fifteen years of commentating on football, I am used to data being wrong. A metric miscounted, a move attributed to the wrong player, a goal credited to someone else. When it is wrong, you fix it. But what I saw that night was different in nature. It was not wrong. It was empty. And in modern football, a gap in the data gets treated exactly like a finished conclusion: people read 'no information' as 'no problem', and read 'the system is dead' as 'everything is fine'. That is the most dangerous con of the analytics era, and almost nobody is talking about it.

Context: from the eye to the chip

In the early 2000s, when I was still fumbling my way through articles for a sports paper in Vietnam, football analysis was the business of the eye and memory. You watched the match, you remembered the moves, and you retold them. The only evidence was videotape and personal credibility. Arguing about football back then was arguing about memory, and everyone had the right to claim theirs was right.

Then everything changed. Opta, StatsBomb, and hundreds of smaller data providers turned every pass, every off-ball run, every shot into a score. By the mid-2010s, xG, expected goals, had become the common language of the analytics crowd. By the 2026 World Cup, every broadcast on the planet carried at least one chart. By the 2026 World Cup in Qatar, a hot take without numbers behind it barely existed.

I understand why football went this way. Data gives an argument an anchor. It turns a vague feeling into a point of reference. When I said Germany would be eliminated at the 2026 World Cup because they lacked a true striker after Miroslav Klose retired, I was using my eye. But with more data on chances created from crosses, my claim would have been harder to dismiss as lunacy.

The problem lies elsewhere: we have built an entire house on a data foundation without really checking whether that foundation is solid. Modern football generates millions of data points every matchday, yet the number of people who genuinely understand how that data is produced can be counted on one hand. Most of us are end consumers, mouths open, catching numbers that fall from a long supply chain that nobody actually sees.

The data supply chain nobody sees

Picture a pass in midfield in a Premier League match. The moment the ball leaves the player's foot, it is tracked by at least two camera systems. A data-collection company turns that information into a digital event. The event is tagged, filtered, cross-checked, and encoded into a data point. The data point travels through an API, is ingested by a third-party provider, and is displayed in an app. A journalist opens the app, screenshots it, and posts it on social media. You and I read it, nod, and quote it again in an argument at a coffee shop.

Along that whole chain, there are at least seven places where an error can occur. A camera can lose its angle. A tagging algorithm can misidentify a player. A human checker can fall asleep. An API can return stale data. An app can display the wrong unit. A journalist can read the wrong column. And the final reader, which is us, has no way whatsoever of knowing whether the data point on the screen passed through all seven stations without distortion.

The most frightening thing is not the error. It is the kind of error that makes no sound.

Silent failure: when the system dies without an alarm

In software engineering, this is called a silent failure. A system fails but still returns a result that looks valid. No error message, no red light, no bell. It simply stops working and returns a gap beautifully formatted as an answer.

In football, silent failures are everywhere, and they are far more dangerous than visible bad data.

When a number is wrong, someone always finds out, sooner or later. Fans watch the match, record it, and cross-check. Rival analysts have an incentive to catch errors. An obvious mistake gets called out, gets fixed, sparks a small controversy, and then dies down.

But when the data is empty, nobody can catch anything. An empty table does not say Team A played well or badly. It says nothing at all. And when an analyst opens that table, their eyes will automatically fill the gap with the bias they already hold. If they already think Team A played well, the empty table gets read as nothing to worry about. If they already think Team A played badly, it gets read as so catastrophic there was nothing left to count.

Empty data is not neutral. It is a mirror reflecting what people want to believe. That is the line I want everyone to pin to their mind, because it explains almost the entire disease of today's football analytics industry.

Three levels of data deception

After years of wrestling with football data, I split the deception into three levels, from easy to impossible to detect.

Level one is wrong data. This is the most benign kind. A shot miscounted as a goal. An assist credited to the wrong player. It gets exposed fast, and the football community has an effective self-correction mechanism: thousands of eyes on a single data point will drag an error into the light within hours.

Level two is incomplete data. A metric is published but stripped of context. A high pass-completion rate can signal an excellent midfielder, or it can simply signal a player who only passes sideways and backwards. The number is right but the story is wrong. This is harder to catch, because it demands tactical knowledge from the reader to see the hidden part.

Level three is empty data. This is the most dangerous, and the least discussed. Empty data is not wrong and not missing context; it simply does not exist. The system failed, but failed quietly. And because there is nothing to refute and nothing to call out, empty data sails through every layer of quality control as a default truth.

These three levels are not separate. They overlap in a familiar sequence: a small technical fault creates wrong data, the wrong data gets trimmed into incomplete data, and then, when the system is exhausted, it abandons the task entirely, leaving behind a gap that is presented as if nothing happened.

xG and the trap of the single metric

Expected goals is the perfect example of everything I just described.

In principle, xG measures the quality of chances a team creates and concedes. A team that shoots a lot but has low xG is shooting from distance, from narrow angles, or under heavy marking. A team that shoots little but has high xG is creating genuinely dangerous chances. In theory, it is a brilliant tool for judging process instead of results.

In practice, xG depends entirely on the model behind it. Each provider has its own model, assigning different weights to distance, angle, goalkeeper position, number of defenders, and even the finishing quality of the shooter. The result is that, for the same match, two providers can produce xG figures that differ by up to three-tenths. To the end viewer, that hardly matters. But to a coach hunting a few percentage points of difference to decide whether to change tactics, three-tenths is an entirely different reality.

And when the xG model meets a kind of match it was never trained on, such as a heavily interrupted game, or a team playing an unusual five-man defence, it can return meaningless numbers. But because they look valid, they are used as if nothing is strange. Nobody reruns the model. Nobody cross-checks. And so an empty number walks straight into a real tactical decision.

The times I almost got it wrong because the data was empty

When you work as a commentator, you learn the most from the times you almost fell. I have had a few.

The summer of 2026 taught me this: people remember the shock merchant more than the contract. I wrote a long piece arguing that Manchester United had thrown a fortune at Romelu Lukaku and should have bought Alexandre Lacazette instead. The two had almost identical numbers the previous season, but the context was completely different: Lukaku played for Everton in a counter-attacking side, while Lacazette played for Lyon in Ligue 1 against weaker opposition. I used raw goal data to compare them and ignored the context. That is a textbook level-two error.

At the 2026 World Cup, I said Germany would be eliminated when the whole world was still dreaming. History belongs to those who speak first. I argued that they lacked a genuine striker in the post-Klose era, and history sided with me when Germany lost 0-1 to Mexico and then 0-2 to South Korea, crashing out in the group stage. But I nearly ruined that call by citing a metric I could not verify. Only later did I discover that the figure I planned to use came from an unsourced aggregator, and it had not been updated after qualifying.

Mispronouncing a player's name is an ear error. Getting a prediction wrong is a career error. I only ever made ear errors. But if I had cited a wrong number, that would have been a genuine career error, and no amount of humour could have fixed it. I still remember the Germany-Mexico match in Moscow, when I called Leon Goretzka 'Gomez' three times in a row on live air. Social media mocked me for a month. I had to review the tape and build my own pronunciation sheet. But I am not ashamed of that. I am ashamed when I think about how close I came to quoting a number that did not exist.

The 2026 pandemic taught me another lesson. When the ball stopped rolling in the pandemic, I asked myself: do I love football, or do I love the feeling of being heard? When the Premier League paused in March 2026, I lost almost all my income from commentary work. I shifted to hosting classic-match rewatches on Twitter Spaces. It was during that period that I clashed fiercely with a data analyst over whether Liverpool's title, won seven rounds early in a shortened season, was deserved. I said a title is a title, stop using numbers to spoil the magic. I was dismissed for lacking evidence, and I realised I needed at least a few figures to hold my own against the bookish crowd. That was the first time I understood that a great talker without data will eventually be left behind.

At Euro 2026, the final between England and Italy took place right at Wembley, a few miles from my London home. I joined the crowd singing Football's Coming Home in the street, but then sat down and wrote a piece headlined that Italy would win because they played with joy while England were afraid of themselves. The piece was shared more than fifty thousand times. Italy won 3-2 on penalties, and I was hailed as a prophet.

But in that piece, I wrote that Italy had scored nine goals in the group stage, when the real number was seven. An Italy fan pointed out the error. At first I joked that I write with my heart and numbers are other people's business. But exactly one week later, I hired an assistant whose job was to check figures before I published. Not because I feared losing credibility, but because I understood that a correct hot take can be pulled down by a single wrong digit.

Then came the 2026 World Cup. On November 22, in Qatar, I watched Argentina lose 1-0 to Saudi Arabia. In the rush to start a trend, I went on air and said: Messi is thirty-five, too old to carry the national team; Argentina will go out in the group stage. The piece exploded. When Argentina lifted the trophy, I became the laughing stock of the football world.

I was forced to write a correction headlined that I had been wrong about Messi, and that this was the most beautiful thing about football. That piece spread more than a hundred thousand shares, thanks to its honesty. But the lesson was not merely to look beyond a single match. It was that behind every one of my claims sits a data chain I had never truly traced to its root.

Vietnamese football and the trap of young data

Here I have to speak plainly about my home market.

Vietnamese football is entering the data era at breakneck speed, but on a young foundation. Each V.League season has hundreds of matches, and more and more statistics platforms are appearing. Clubs are starting to hire analysts. Match reports are starting to carry figures. That is a good thing.

But when I look at how data is used in Vietnam, I see a familiar problem: most of the numbers come from abroad, from providers nobody can verify, and they are used without context. A newspaper cites an international provider's metric without knowing that the provider does not fully cover V.League. A coach uses GPS data to assess a player without understanding that the device can lose signal in a small stadium. A fan quotes V.League xG without knowing that the figure was produced by a model built for European leagues.

I remember a trip home, sitting over coffee with a friend who does analysis for a V.League club. He told me that most of his time goes into checking data, not analysing it. Pass data is often off because the camera angles do not cover the whole pitch. Distance-covered data is often wrong because GPS devices drop signal when players move under a roof. And the line I will never forget: here, we have to teach the system what a match is before we can teach it which player is good.

That is the Vietnamese edition of the silent failure. We consume data without having built a verification system. We trust numbers produced elsewhere, for other leagues, by other standards. A young football culture as emotional and distinctive as Vietnam risks losing that very identity if it lets foreign numbers define it.

And I say this as a Vietnamese man working in football in England, not as an outsider pointing fingers. I want to see V.League with its own data system, built by Vietnamese people, understanding Vietnamese pitches, understanding V.League's tempo, rather than importing a metric set built for the Premier League and pasting it onto Vietnamese players.

Who verifies the verifiers?

In football there is a question few dare to ask: if all data comes from a long supply chain, who is the last person guaranteeing it is right?

In many major markets, independent sports-data auditors exist. But in most leagues, including the biggest, verification still rests on trust. Provider A says this metric is correct. Provider B says that metric is correct. Nobody has the authority to judge. And when two providers give two different figures for the same match, fans usually never know, because platforms only pick one.

This creates a paradox: the more data, the less verifiability. The less verifiability, the more gaps. And gaps are where silent failures breed.

On reflection: maybe I am wrong

I have to admit something before I finish. There is a powerful counterargument to my case, and I do not want to pretend it does not exist.

The counterargument goes like this: data, even flawed, is still better than the eye. Before the analytics era, football was drenched in bias, regional favouritism, and baseless myths. Data itself has helped uncover undervalued players, effective tactics that were overlooked, and unspoken prejudice about race, nationality, and physique. If forced to choose between an imperfect data system and a system based purely on feeling, data still wins.

I agree with most of that counterargument. But it does not refute my point. Data beats the eye only when the data exists. When the data is empty, we do not fall back on the eye; we fall back on bias, now wrapped in a fake coat of science. An analyst opens an empty table and believes they are reading data, when in reality they are reading themselves.

The biggest blind spot in the football analytics community is that we have invested heavily in learning how to read data, but almost nothing in learning how to recognise when data does not exist. We can read an xG chart in three seconds, but we cannot notice that the chart has not been updated in two weeks. We can argue about a metric for three hours, but cannot answer a simple question: where did this number come from, and who last checked it?

A correct hot take does not make a name. A wrong hot take, at the right moment, becomes a legend. But a hot take built on empty data is not a legend. It is an accident waiting to explode.

What I believe will happen

I will stake one verifiable prediction: within three years, a major data scandal will erupt in a top league. Not a match-fixing scandal. Not a betting scandal. It will be one in which thousands of tactical decisions and millions of fan opinions were built on a data source that died long ago without anyone noticing.

When that happens, football will face a question bigger than which team wins the title: do we truly understand the tools we are using? And if the answer is no, then what exactly is our confidence built on?

At forty-two, I still write as if every match is the last one I will ever live. But now, before I write a single line, I open the data table and check the first thing: does it actually have anything to say? Because in modern football, the most dangerous enemy is not the liar. It is the silence we mistake for agreement.