The Verdict Without Evidence: Mike Dean's 47 Decisions and the Biggest Hole in Football Analysis

Anfield, February 11, 2026, minute 73. The score was 1-1 and Sadio Mané ran onto a ball played through the line in a position the Kop could read more clearly than the assistant referee. The flag stayed down. The goal stood, the match ended level, and within twelve minutes of the final whistle social media had its verdict ready: Mike Dean had ruined another game.

I stayed in my seat for another forty minutes. Not out of anger. I opened my notebook, drew a table, and started counting. That night I logged all 47 decisions made by the officiating team — every free kick outside the box, every advantage played, every position taken by the referee, every moment the assistant raised the flag. When I checked the log against the broadcast footage, the result made me read it twice: Mike Dean was correct on 46 of 47. Wrong exactly once, and that single error decided the match.

Nobody remembers the forty-six correct calls. The whole country remembers the one mistake. That is the nature of officiating, and it is also the nature of every football argument today: we judge by the most visible incident, not by the base of evidence underneath it. Football analysis has a hole bigger than any VAR error, bigger than any contested penalty, bigger than any offside-law debate. The hole is this: most verdicts are issued with no evidentiary base at all, and nobody owns the job of building that base.

When data walks into the dressing room, emotion has to leave through the window. But before data gets there, the dressing room is already empty. That is the problem.

Context: an industry of judgment with no baseline

Over the past decade, European football has undergone three infrastructure shifts at once. VAR entered the Premier League in the 2026-20 season. Semi-automated offside technology arrived at the 2026 World Cup and was later adopted in the English top flight. Advanced performance metrics — xG, xGA, PPDA, touches inside the box — became standard vocabulary for recruitment departments and studio panels alike.

The paradox is that referee data infrastructure and media data infrastructure have grown at different speeds. Referees gained tools but no comparative baseline. Broadcasters gained numbers but mostly use them to decorate conclusions already written. The result is a gap in the middle, where verdicts are announced without a reference point and fans are left clinging to feeling.

I used to be a VAR sceptic, and that is exactly why I understand the people who hate it. In 2026, when fellow commentators in London lined up to attack the disruption VAR caused, I did not argue with emotion. I started a stopwatch.

Core: four counts, and what they taught me

The first count was at Anfield in 2026: 47 decisions, measured against twelve criteria I built myself — referee position, sightline obstruction, distance to the incident, reaction time, assistant involvement, ball direction, speed of play, contact posture, plus four contextual variables. The spreadsheet was nothing fancy. It did one thing the media does not do: separate the salient from the aggregate.

The result changed how I write. Mike Dean's error that night was what I call a cognitive error, not a technical one. A technical error is standing in the wrong place and missing a challenge. A cognitive error is standing in the right place, looking the right way, and reading the situation wrongly inside 0.3 seconds. The two need entirely different remedies. One is fixed by positioning drills. The other needs tools. Merging them into a single phrase — "bad referee" — fixes nothing.

The second count was at the 2026 World Cup. On June 16, 2026, France played Australia and Antoine Griezmann opened the scoring from the penalty spot after a VAR intervention. Colleagues in the studio called it the end of football's rhythm. I sat in the BBC edit room, timed every review, and logged the final number. The average review length I measured was 101 seconds. I totalled the other 14 VAR decisions across the tournament and found added time increased by an average of 2 minutes 37 seconds per match.

Two minutes thirty-seven. That was the entire price. Meanwhile a throw-in in the 88th minute routinely eats more time than that, and nobody campaigns to abolish throw-ins.

I did not conclude that VAR is a good tool. I concluded that the argument "VAR destroys the rhythm" is an argument with no data behind it, and an argument with no data behind it cannot be rebutted and cannot be fixed. It can only be replaced by another argument with no data. That is the loop English football has been stuck in for seven years.

The third count was the summer of 2026. When the Premier League returned after lockdown, I contributed analysis to an independent study on crowd influence over refereeing decisions. I collected data from 89 matches before and after the pandemic. Two metrics moved sharply: yellow cards fell 23 per cent, and penalties rose 31 per cent in empty stadiums.

The popular reading is that crowds pressure referees, so with empty stands, referees reach for cards less. That reading is correct but incomplete. The more telling shift runs the other way: penalties went up. Without 50,000 voices behind them, referees became more decisive on box judgements that carry a punitive character, and more cautious on sanctions that manage the game. Crowd pressure does not make referees wrong. It shifts the weighting between two categories of decision.

I kept those findings in a drawer for four months. Perfectionism made me verify the dataset three times and cross-check with two analyst colleagues. When the piece finally published, UEFA data analysts cited it. But those four months cost the data its currency. An empty stadium does not lose its soul; it returns the soul to its rightful owner, and the owner is the game itself, which runs whether or not anyone is watching.

The fourth count was at the 2026 World Cup. On November 21, 2026, England beat Iran 6-2. The stands talked about Bukayo Saka. I sat in the VIP section and watched Jude Bellingham, then nineteen years old. I logged every touch. By full time I had 78 touches, 41 of them first-time, and almost none where he held the ball longer than three seconds.

Three seconds. That is the fingerprint of a modern central midfielder. Not assists, not goals. The capacity to decide before the ball arrives. I called a former scout to cross-check, and he confirmed what I had seen. After the tournament I wrote a series on Bellingham and predicted he would become the best central midfielder of his generation, before any major outlet put him on a front page.

Four counts, four lessons, one methodological conclusion. Macro statistics tell you who won. Micro statistics tell you why. Only the second kind lets you see what is about to happen rather than what already has.

Three kinds of evidence football is missing

If I had to name exactly what is absent from most Premier League refereeing debates, I would name three things.

A seasonal baseline. The laws change almost every year, yet no governing body publishes a baseline table showing the real-world effect of each change. The handball law is the clearest case. From 2026 the interpretation of "unnatural arm position" was tightened. By 2026 guidance was loosened and accidental handball by a teammate no longer cancelled a goal in the build-up. Every rewording forces millions of fans to rebuild their intuition from scratch, with no reference point supplied.

Accepted error margins. Aviation publishes the failure probability of each component. Medicine publishes false-positive rates for each test. Football publishes no error margin for semi-automated offside technology, despite its use at the 2026 World Cup and subsequently in the Premier League. Without a margin, every dispute becomes a dispute about belief, and belief cannot be audited.

Process data. We see the final decision, not the deliberation. We do not know how many camera angles the VAR team reviewed, in what order, how long each took, which criteria came first. Cameras find the error, but only people find the cause. Publishing the process would not weaken referees. It would make them less misunderstood.

None of the three requires new technology. All three require an administrative decision: to publish. And that cost is precisely what leagues avoid, because publishing a baseline means tying your own hands in front of public opinion.

Counterintuitive angle: VAR is not football's enemy

The anti-VAR side has a real feeling. The data side has a real argument. Both are defending the right thing with the wrong instrument, which is why this debate has not moved in seven years.

The anti-VAR camp says technology kills the moment. They are right. The explosion that follows a goal is split into two stages, and no technology gives that wholeness back. But they misidentify the culprit. What kills the moment is placing an incident in the hands of three people in a sealed room in London, where nobody knows what they are watching, what they are weighing, or when they will conclude. Ambiguity kills the moment. Technology merely amplifies that ambiguity.

The data camp says numbers will settle disputes. They are wrong too. Numbers do not settle disputes. They relocate the dispute from "what do you think" to "how do you define it". Handball does not divide people because of missing data. It divides people because everyone holds a different definition of "unnatural position", and no definition has been fixed in wording concrete enough to apply.

The place where I have to turn the mirror on myself, after thirty years in this industry, is a period when I trusted data too much. Back in 2026, having measured World Cup review times, I wrote that if each intervention cost 101 seconds and each match gained only 2 minutes 37, VAR was essentially free. That argument ignored a variable I never measured: the emotional cost of waiting. A supporter waiting in a state of not knowing whether they are allowed to celebrate loses something no metric captures. I was right about the data and wrong about the people. Since then, every analysis I write must carry one line for the unmeasurable variable.

The best referee is the one nobody mentions after the match. But in a system where the process itself is sealed, the person nobody mentions is the one most easily suspected.

What will change, and what will not

Semi-automated offside has solved part of the problem. It has cut the time needed to establish offside position to a level nobody thought possible, and for objectively offside cases it has almost eliminated dispute. But it cannot solve the hardest part: the interval between the technology producing an answer and a human deciding to publish it. Every remaining football controversy lives inside that interval.

A referee's authority does not come from the whistle; it comes from reading the situation. The right reform is not to hand more power to technology, nor to hand everything back to referees. The right reform is to force both sides to publish the evidence base they rely on.

One concrete proposal, and I know it will be refused for commercial reasons: publish a seasonal baseline table for each category of refereeing decision, with accepted error margins and a classification separating technical error from cognitive error. The Premier League already holds this data through its independent expert panels. It simply chooses not to publish.

The most valuable contract is usually the one never announced. In this case, that contract is the refereeing dataset — an asset leagues already own and keep sealed for fear it might backfire.

I have spent nearly a decade counting, measuring and logging what others skip. Not to prove referees are good, and not to prove VAR is bad. I did it for one simple reason: when a verdict is issued with no evidence behind it, what suffers is not the person being judged, but the judgment of everyone else in the room.

Vietnamese football stands at exactly this fork. As VAR arrives in V-League and regional competitions, we have the chance to do what Europe failed to do: build the baseline from the start, publish the process from the start, and teach supporters how to read a decision instead of only how to react to one. Otherwise we will import England's seven years of argument wholesale, just a few years late.

Who will be the first to publish their refereeing dataset?