Schema-Complete, Content-Empty: The Silent Disease of Football Analysis

Schema-Complete, Content-Empty: The Silent Disease of Football Analysis

Three in the morning in Tokyo, in the middle of a transfer window that pushes thousands of rumours across the screen every day, I opened a thirty-four page dossier sent by a data analysis firm to a J1 League club. The covering note was one line: "Post-match report complete." It had a title. It had a table of contents. It had a heat map. It had an xG chart. It had tactical recommendations for the next matchday. Structurally, not a single box was missing.

I read all thirty-four pages and learned nothing about the match.

In more than four decades in this trade I have learned to separate two kinds of document. The wrong one can be checked, argued with, corrected. The other is correct in every check box and empty in every conclusion. It has the shape of knowledge, the smell of knowledge, and no knowledge inside it. That second kind is more dangerous, because nobody knows where to begin the argument. At sixty-one, I have no time left for the polite version of football on paper.

Football industrialised its analysis layer over roughly fifteen years, and the transfer window is where that process shows its true face. Every top-flight European club now employs between three and fifteen data staff. Every sports desk keeps an xG dashboard open in a browser tab. Every account with nine hundred followers can build a comparison table in four minutes. Analytical output is growing exponentially. The supply of new signal is flat.

What I see in this window is not a shortage of data. We are drowning in the stuff. The problem is that most of it is packaged into a format that was completed before anyone asked a question about the content. Nobody builds the report around a finding. People build the report around a template and then force whatever fits into it. The template is never empty. The template is only empty of information.

I borrow a term from data engineers for this phenomenon: null propagation. In a data pipeline, when a value is missing or empty at the input stage, the system usually does not throw an error. It keeps running. It carries the empty value through every subsequent calculation, keeps the structure, keeps the output format, and emits a result that looks finished. No exclamation mark appears. No red screen lights up. There is only a beautiful and meaningless report.

Football is running exactly like that pipeline. Three production lines generate the most empty reports: the post-match verdict, the transfer valuation model, and the daily tactical commentary. All three share one property — they are designed to always have something to publish, regardless of whether anything was found.

Try the simple test I have applied to every report that has reached my hands over the past two years. Read the concluding sentence, then ask yourself: could this sentence have been written about a different match? If the answer is yes, the report is empty, however handsome its formatting.

Take one sentence I encountered four times in a single week, in four different leagues, written by four different analytics providers: "The home side controlled possession well but lacked ideas in the final third." That sentence is true of roughly sixty per cent of all football matches ever played. It is not wrong. It is useless.

This is where I have to retell the case that shaped how I have written for the past seven years. The 2026 World Cup, Spain against Russia in the round of sixteen. I sat in front of a screen in Tokyo, watching and pulling live data. Spain held around 74 per cent of possession, completed upwards of a thousand passes, and finished the match with roughly 0.9 xG. Russia defended in a low 5-3-2 block, barely attempted to control the ball, and scored from a set piece. The match went to penalties, and Spain went out.

What I wrote on social media that night was one line: "Possession is an illusion; pressing and transitions are real." Forty-seven traditional journalists attacked me for it. Six hours later the post had two thousand three hundred shares.

The share count matters less than the structure of the report I had implicitly demolished. The standard report on Spain versus Russia has every box: possession, pass count, pass accuracy, entries into the box, shot volume. All of it correct. All of it empty, because not one box answers the only question of value: how does this team score, and how does the opponent remove that ability?

Tiki-taka did not die because it was beaten; it died because it was trusted for too long. And it was trusted for so long precisely because the reports describing it were always complete in format and always hollow in substance. When every metric looks good, nobody has a reason left to ask what the metrics mean.

The pandemic of 2026 gave me a laboratory no university could have built. When the Bundesliga returned in May 2026 with empty stadiums, I began logging every match. Eighty-seven matches ended up in my sample. The home win rate fell from 43 per cent to 31 per cent. The draw rate rose to 29 per cent. The home advantage margin was cut by nearly half simply by removing the sound of people.

The empty stadium of 2026 was the laboratory; only now are we seeing the finished product.

There is one detail in that dataset I consider the most important, and almost nobody mentions it. Possession figures barely moved. Passing figures barely moved. xG figures barely moved. What moved was the result. In other words, hundreds of reports produced during that period kept their old format, stayed correct in every box, while the variable that decided the outcome of matches had disappeared from the system without a single metric registering it.

That is the most precise definition of null propagation in football. The system did not lose its data. The system lost its meaning, and not one row in the reporting table was designed to detect that loss.

In the transfer market, the disease takes its most dangerous form, because here it spends real money. Modern transfer valuation models are built around predictive indicators of potential: minutes played under the age of twenty-three, expected threat, season-on-season improvement rate, projected resale value. All of that can be measured, and all of it enters the model with explicit weightings.

What has no column in the model is the dressing room.

I have tracked enough deals over fifteen years to see a repeating pattern: a club pays a record fee for a nineteen-year-old with a beautiful improvement curve, drops him into a squad whose wage structure has been stable for seven seasons, and breaks that structure within the first week. The transfer report is still schema-complete. It has every box. It has no box for the question: how will the thirty-two-year-old captain react to a newcomer earning three times his salary?

This is why transfer data models overvalue young potential and undervalue dressing-room chemistry. This is not a defect that can be fixed by adding a variable. It is a structural defect: most of what makes a signing succeed lies outside every column the model was designed to read.

Japan gave me an angle I never had while working in England. Japanese football values system, patience and process. Those are genuine qualities, and they have produced one of the best-organised football nations in Asia. But precisely because it prizes process, Japanese football absorbed template-reporting culture faster and deeper than anywhere I have worked. Here, a report is judged first by its compliance with the template.

The phrase I hear most often in tactical meetings in Tokyo is "according to plan". The team lost, but according to plan. The team created nothing in the second half, but according to plan. The team conceded from a set piece the plan never accounted for, and still according to plan. I have nothing against a plan. I object to the plan becoming a substitute for judgement. When process is venerated to the point that nobody dares ask whether the process still describes reality, the process has become an organised form of empty reporting.

For four years now I have brought one more lens to this trade, and it makes a lot of my European colleagues uncomfortable. I study the analytical language of esports. There, people talk about rotation, power spikes, map control over time, and the dead window of a composition before it reaches its strength threshold. These are concepts that describe the movement of advantage in real time.

I declared in 2026 that esports is the modern Olympics. The IOC laughed. Now they are chasing us.

What I learned from that lens is not that esports is superior to football. It is that esports analysts are forced to answer the question football analysts are permitted to dodge: where is the advantage right now, and when does it expire? An esports breakdown cannot end with "the home side controlled well but lacked a cutting edge". Nobody pays for that sentence.

That is the standard I want applied to football. Not a standard of more numbers, but a standard of a conclusion that can be proven wrong.

And here I have to write the part that people who have followed me for years know I always write, even though it weakens my own case.

I could be wrong in three places.

First, templates may be necessary scaffolding. A twenty-five-year-old analyst who has not watched three hundred matches live needs a structure to begin with, and an empty structure is still better than no structure. If I deny that, I am demanding of others something I only earned through forty-five years of watching.

Second, the market may filter this out on its own. An empty report does not win matches, and a club that pays for one will learn that within a few seasons. If that self-correction mechanism exists, the disease I am describing is a transitional phase rather than a systemic fault.

Third, and this is where I doubt myself most, I may have produced an empty report inside this very article. In 2026 I drew a large conclusion about possession from a sample of one match. One match. I did exactly what I am criticising: took an existing format, dropped one data point into it, and published. If so, this piece deserves the same test I propose: could its conclusion be used for a different subject?

I still believe it could not. But I leave the judgement to the reader, and I leave behind a measurable claim.

Over the next twelve months I will collect one hundred post-match analyses published on major sports platforms in Asia and Europe and count how many contain at least one finding that cannot be derived from the scoreline. My prediction: fewer than twenty. I will publish the full list, the marking criteria, and the ones I consider excellent, so that anyone can check the work.

If the real number comes in higher than my prediction, I will be the first to write that I was wrong and that this industry is healthier than I believed. If it comes in lower, then the problem is no longer the opinion of one difficult commentator in Tokyo, but an industry feeding itself on reports that contain no information.

Football has never lacked data. What it lacks is people willing to close the report and say there is nothing inside.