When the Data File Is Empty: A Football Analyst Learns to Stay Silent

On January 31, 2026, I opened a file on my screen and found it empty. Not a single line of metrics, not a season, not even a name. The report carried exactly one question: was Enzo Fernández worth the 106.8 million pounds Chelsea were about to pay Benfica? I sat in front of that screen for nearly an hour. Outside the window, Liverpool in January was cold and grey. At that point Enzo had spent only half a season in European football, and the volume of reliable data needed to build a model barely existed.

If you have heard me talk about xG, about PPDA, about ball-circulation speed, you might think everything comes down to numbers. That is half true. The other half — the harder half — is learning to recognise when the data is not enough to say anything at all.

My work runs on two layers. The first layer breaks a season down into discrete units of information: team names, player names, minutes played, top-flight appearances, sources, publication dates. The second layer takes those units and builds a judgement. After thirty-five years, I have settled on one discipline: the second layer must never run while the first layer is empty. A report that carries only the label "football" and no content is not yet a report; it is a silence. And a silence, if filled with guesswork, becomes a polite lie.

The answer I had to give the club that day was simple: not enough data. Thirty top-flight appearances are not enough to price a player at a quarter of a big club's transfer budget. People do not pay to hear "I don't know". They pay to hear a prediction, and the more decisive the better.

But the market itself tells a different story. In January 2026, Chelsea paid 106.8 million pounds for Enzo Fernández and 62 million pounds — potentially rising to 89 million — for Mykhailo Mudryk, who at the time had played only a handful of Champions League games. In August of the same year, they paid 115 million pounds for Moisés Caicedo. Three deals, nearly 300 million pounds, most of it resting on samples so thin they can hardly be called samples. In statistics, that is the worst thing you can do: draw a conclusion from a few data points and call it a trend.

I once stood before a spreadsheet and felt as though I were witnessing a miracle at Anfield. But I also learned that such belief only holds once everything unverifiable has been stripped away. Faith in data without humility is just a new form of superstition.

Consider xG, the metric I have pursued for ten years. xG measures the quality of chances, not the result. A team averaging 2.4 xG per match is a team creating many high-value chances. But xG will not tell you who will score, who will get injured, who will lose form after a summer. xG speaks about process, goals speak about outcome, and the gap between them is where the story of football begins.

At the 2026 World Cup, I built an xG model for all 64 matches and predicted France would win from the group stage. The model was right, but for a different reason than I assumed. I had argued Croatia were winning on luck, then spent two weeks in a library rewatching every passage and discovered I had omitted set pieces entirely. My model was not wrong because it mispredicted. It was wrong because it stayed silent about part of the truth. Since then, I never publish a conclusion without stating what it lacks.

xG is a revolution, but every revolution needs time before people accept it. The same applies to pressing advocates like Klopp: only after he won did people stop calling his style suicidal. Those who are right before their time always pay for it in solitude.

The hardest part of this job is finding the ground the data has never walked on, and speaking plainly about the emptiness there. An empty spreadsheet is not an excuse. It is a complete and correct conclusion. When there is no data, "I don't know" is the most honest judgement available.

But football does not reward silence. It rewards noise. Deadline day exists to sell emotion, not to communicate truth. In that environment, the person who says "I don't know" is seen as lacking nerve. That is why most of the polished transfer reports on the market are built from the desire to have a story to tell, rather than from data.

There is a subtler trap than the temptation of prediction: when a system raises no risk flag at all, people easily assume it is clean. The absence of a red flag, in the worst case, is simply an unchecked gap. That is why I force my spreadsheets to distinguish clearly between "safe" and "unknown". The two look identical on screen, yet they differ enormously in reality.

The same logic applies to refereeing. When VAR arrived, people promised mistakes would be eliminated. But VAR's intervention threshold is "clear and obvious error". That phrase is itself an ambiguous field, because what is clear to one person may be murky to another. Technology does not erase subjective judgement; sometimes it merely moves that judgement from the on-field referee to the VAR room and repackages it in technical language.

In a world of endlessly long seasons, the awakened can only rely on their own spreadsheet. Mine, on January 31, 2026, was empty. And I said exactly what it forced me to say. The winter 2026 window saw Chelsea spend over 300 million pounds in a single month, mostly on players with fewer than 30 top-flight appearances. If a similar wave returns next season, perhaps what we need first is not another prediction, but a spreadsheet that can withstand blank space — and a person brave enough to leave it blank.