A Complete Report With No Content: Why Modern Football Analysis Is Fooling Itself

A Sunday night in April 2026, in the video room of the Hamburger SV youth academy. I was thirty-five, and in front of me lay forty-seven VHS tapes of the under-19 side's 2026-98 season, along with a sixty-page dossier written in my own hand. Every page was packed with filled-in boxes: starting shape, pass completion, duels contested, high-pressing minutes, turnovers split by thirds of the pitch.

Not a single box was left empty. And not a single line answered the question I was chasing: why did this team lose 73 percent of its matches whenever it faced an opponent lined up 3-5-2 with two holding midfielders.

I sat there turning the pages for a long time. The dossier was perfect in form and empty in content. It recorded everything that had happened and explained nothing.

From the HSV video room, I see the Bundesliga as a chessboard. But it would take another twenty years, and the season of football without crowds, before I understood the true scale of the disease I had brushed against that Sunday night.

In the twenty-five years since, the football analysis industry has moved in the opposite direction from what I learned. It has become more complete, more standardised, and emptier.

A single Bundesliga match today generates a volume of data that a video analyst in 2026 could not have imagined: thousands of tagged events, the coordinates of every pass, spatial density models, expected goals, expected assists, passes allowed per defensive action, field tilt. Data platforms sell clubs automated reports, every box filled, correctly formatted, printable the moment the final whistle blows.

The problem lies elsewhere entirely. An automated report cannot say "I do not know". It has no empty box to leave empty. When the input data is corrupted — a clip that fails to load, a match where the recognition system tagged only half the events, a truncated record — the machine does not stop. It still prints a document with nine complete sections, charts included, conclusions included.

That is when the empty report is born. Not a wrong report. A wrong report is useful, because you can argue with it. The empty report is far more dangerous: correct in form, fluent in language, and containing not a single bit of information that can be used to make a decision.

The crux is this: a document with no formatting errors can still be entirely worthless, and modern football is producing that kind of document at industrial speed.

The coaching staff read it, nod, fold it away, and walk into the next match believing they have been equipped. Nobody notices that the entire analytical section is an empty skeleton wrapped in terminology.

World Cup 2026 was not a tournament. It was a tactical case file. And inside that case file, I saw the archetype of this disease in one specific match.

On 16 June 2026, in Kazan, France played Australia. The final score was 2-1. I had been assigned to Group C and spent two full days on this match, building a spatial density map in fifteen-minute blocks.

The automated report for that match, if printed, would look flawless. France dominated possession, took three times as many shots, overwhelmed the final third with passes, and posted a clearly superior expected goals figure. Every box was filled. The conclusion would read: France imposed their game, deserved the win, Australia defended negatively.

No box in that report recorded what I saw on the tape.

Australia's block sat so deep that the lowest defensive line stood nineteen metres from the goal line. Nineteen metres. That is a distance professional football at this level has almost stopped producing, because it turns the penalty area into a corridor with no depth to exploit. Antoine Griezmann won a penalty awarded by the referee through video technology — the first penalty in World Cup history given that way. Mile Jedinak equalised from twelve yards after Samuel Umtiti handled. Paul Pogba settled it with a shot that deflected.

All three goals came from something other than open play. They came from a penalty, a penalty, and a deflection. The entire volume of possession France generated produced no goal from organised open play.

This is what the empty report conceals. It does not lie. It simply does not ask. It does not ask: why did a team with that much possession fail to create a single clear chance from its attacking structure? It does not ask: what problem did a block at nineteen metres create for the tempo of the first pass after recovery?

The answer lay in the fact that France recovered the ball constantly but released it too slowly. Each time they won it back, the midfield took an average of nearly two and a half seconds to move the ball past the halfway line. Against a block sitting at nineteen metres, two and a half seconds is more than enough for Australia's two banks to close into a dense four-one-four-one. France were not stopped by the defence. France were stopped by their own processing time.

That is a verifiable conclusion. It points to a specific adjustment in the next training session: reduce the number of touches by the central midfielder before releasing, focus on the first pass played through the lines. The automated report can never produce that, because it counts passes; it does not measure the interval between winning the ball and moving it past the halfway line.

Miracles on the pitch are only calculations the crowd has not yet read. But the person writing the report had not read them either, because his spreadsheet had no column for time.

Two years later I got a second chance — and this time it was a genuine laboratory.

In May 2026, the Bundesliga returned in empty stadiums. I was fifty-seven, and I recognised immediately that this was a research condition modern football had never offered: one variable removed from the system while almost every other variable stayed fixed.

I analysed eighty-nine matches played behind closed doors in the 2026-20 season. Three figures stood out. Home advantage almost vanished — the home win rate fell to roughly the level of away teams. Pressing intensity dropped 8.3 percent. And pass accuracy rose 3.2 percent.

With the stands empty, tactics stand exposed as if under a microscope.

The interesting part lies in how those three figures are explained. Here, the empty report nearly fooled me once again.

Looking only at the 8.3 percent drop in pressing intensity, the obvious conclusion is that teams played more cautiously. But when I cut the tape situation by situation, the real mechanism looked entirely different. Crowd noise, it turned out, is not merely atmosphere. It is a pressure signal. When an opposing midfielder received the ball in the area in front of the box, the roar from the stands was a cue for the three players around him to press simultaneously. That cue disappeared in an empty stadium.

Without a crowd, the defensive midfielder has to read the situation himself. He reads it half a beat slower. And that half beat reduces pressing intensity across the whole system — not because the team is more cautious, but because a collective trigger has been severed.

By the same logic, the 3.2 percent rise in pass accuracy was not better play. They could hear each other. A defender calling a teammate's name from thirty metres away reached the receiver's ears without being swallowed by noise. On-pitch communication changes completely when the acoustic environment changes.

That is the silent football tactical model I built after that season. It could not have emerged from any automated dataset, because automated datasets have no box for "pressure cue from the stands" or "audibility of a teammate's call".

Both cases — France against Australia, and the silent season — lead to the same conclusion about the nature of tactical analysis.

The value of a report does not lie in its completeness. It lies in whether it identifies which variable has not yet been measured.

I see this most clearly in the two metrics the analysis industry abuses most.

Gegenpressing has been decoded. I first argued this several seasons ago and was contested. But watch the numbers of mid-table Bundesliga sides over the last three seasons. Their passes allowed per defensive action have fallen steadily — meaning they press harder by the metric. Yet their points totals have not risen accordingly. And their soft-tissue injury rates have climbed.

What is happening is that mid-table teams are turning football into athletics. They do not press to win the ball in valuable positions. They press in order to press. The metric improves; the table does not.

Passes allowed per defensive action measures effort, not intent. A team that charges forward in disarray and leaves space behind will post a better figure than a team that presses with structure and wins the ball in the opponent's half. The second team is the one controlling the match.

Expected goals is abused in a similar but subtler way. It describes chance quality and does not explain match decisions, player form, or refereeing standards.

Last season I watched a Bundesliga match in which the home side generated 2.1 expected goals and lost 0-2. The automated report would say the home team deserved at least a point. But when I cut the tape, both goals conceded came from second balls after corners, in the same zone, fourteen minutes apart. No expected goals figure records that the home side lost its second-ball defensive organisation at the near post for the entire final twenty minutes.

The report had one high number. It had no specific action to fix on Tuesday.

By contrast, Bayer Leverkusen's 2026-24 season is an example of data that genuinely contained information. The unbeaten run across the whole campaign was not luck, and that can be verified with one narrow observational sample: goals scored in the final fifteen minutes, and substitutions that directly produced goals. That is a structure, not a run of coincidences. But to read it, an analyst must abandon composite metrics and count specific events by match minute.

This is where I want to stop, because it is the industry's biggest blind spot.

In a risk assessment handed to a coaching staff, there is a life-or-death difference between two lines: "low risk" and "not assessed".

They look nearly identical on paper. They lead to two completely opposite decisions.

When a centre-back returns from an anterior cruciate ligament injury and the data department does not yet have a sufficient sample to conclude anything about re-injury probability, the automated report will not leave that box empty. It fills in "low risk", because the available data shows no anomaly. But the available data is three training sessions. No anomaly across three sessions does not mean low risk. It means nobody has measured it.

So the player goes out in his first match back, told to prove himself. That is a cruel demand in motor-science terms. In a comeback match, a player is not only facing an opponent. He is facing a neuromuscular system that has not been reprogrammed, compensatory movement patterns formed during rehabilitation, and psychological pressure to demonstrate that he is still intact.

Not assessed is not the same as low risk. Equating the two is an analytical error, and in this instance it is paid for in ligaments.

Every contract is a gamble, but I prefer to count probabilities. And counting probabilities begins with admitting when you do not have enough of a sample to count.

What worries me is that the mechanism producing the empty report tends to spread. A pipeline broken at the input layer will not stop with one article. It will keep producing documents with the right format, the right terminology, and no content across the entire next batch, until someone takes responsibility for checking.

In football, the person responsible for checking is the head coach, reading the report at eleven at night before matchday and deciding whether to trust it.

I once witnessed a case where an empty report nearly produced a wrong decision in a major match. The report said the opponent defended set pieces man-to-man. On the tape, across their four most recent corners, they had switched to zonal marking with nobody recording it, because the tagging system was still using the old category. Four situations. A small sample, but it is information. The report had no box for it.

We have built an industry capable of describing everything and explaining very little. We measure passes and not intervals. We measure pressing volume and not where the ball is won. We measure expected goals and not second-ball defensive organisation. We fill every box and never leave one empty, because an empty box looks like failure.

But the empty box is where the information lives.

At sixty-three, I no longer chase the ball, only its intent. And its intent usually sits exactly where the spreadsheet has no column.

So before every match from here, I will do one concrete thing. I will ask myself a single question and write it at the top of the page: of everything I know about this match, what have I genuinely not measured?

If the answer is nothing, I will rewatch the third tape. There is always a third tape.

And if my report ends with an empty box clearly marked "insufficient data to assess", I will treat that as the best result of the working day. A coach who reads that line knows exactly what to watch in the first half. A coach who reads "low risk" knows nothing, and walks into the match holding a belief with no foundation.

The next match will verify it. Not by the scoreline, but by whether, after the final whistle, anyone on the coaching staff can say one specific thing about what happened and why it happened.

If nobody can, then we have just watched a football match through an empty report.