The Anfield 2026 Ledger: 47 Decisions, One Mistake, and How Data Pushes Emotion Out the Window

Minute 73 at Anfield, February 2026. Liverpool led Sunderland 1-0 on an afternoon when the hosts controlled possession but never built a comfortable cushion. Sadio Mane received a through ball, starting roughly a metre and a half ahead of the Sunderland back line. The assistant's flag stayed down. Referee Mike Dean awarded the goal. The match finished 1-1, and by the following morning the entire country was talking about a single incident.

I stayed in the stand for another forty minutes. Not to write impressions. I opened a notebook and listed every free kick, every advantage played, every collision waved away, every exchange with an assistant. Forty-seven decisions across ninety minutes plus stoppage time.

At home I built a spreadsheet with twelve criteria: body position, viewing angle, distance to the point of contact, reaction time, degree of obstruction, player pressure, crowd noise, and seven further variables. I cross-referenced everything against footage from six camera angles.

Mike Dean got one of forty-seven decisions wrong. That single mistake decided the result.

CONTEXT

I lost sleep over that ratio for weeks. 97.9 percent accuracy looks handsome on paper. Football does not award points by percentage. It awards them by goals.

Referee analytics barely existed in public form at the time. Broadcasters invited former officials to confirm or deny a single incident. Nobody built a measurement process. Matches were judged by feel, by the roar of the crowd, by how many column inches a decision generated.

I came at the work from the opposite direction. Born in Vietnam, educated in England and settled there, I worked in an environment where every argument was settled by pre-loaded bias: big clubs get favoured, referees fear the crowd, away teams suffer. Those biases are not entirely wrong. They had simply never been tested against a number.

From February 2026 I worked to one rule: every judgement sits behind the data. If I wanted to claim a referee was biased, I had to prove it through the distribution of decisions across match time and scoreline. If I wanted to claim a player was treated unfairly, I had to prove it through the count of unpenalised collisions against the league baseline.

The rule sounds dry. It saved me from being wrong many times.

CORE ANALYSIS

Three studies are the ones I return to most, and all three begin with a very specific question about opportunity cost.

The first was the Anfield spreadsheet. Splitting forty-seven decisions into two groups revealed something commentators never mention: technical error and cognitive error are different species. Technical error occurs when a referee stands in the right place, looks the right way, but the ball moves faster than the human eye can process. Cognitive error occurs when a referee stands in the wrong place, or lets an external variable intrude on the judgement.

Of Mike Dean's forty-seven decisions, four were difficult technical cases, and he handled all four correctly. The mistake in minute 73 belonged to the second group. The assistant was poorly positioned, and the noise inside Anfield at that moment exceeded a threshold that would affect anybody.

The second study took place in Russia, June 2026. The BBC brought me in to analyse VAR across group-stage matches. The entire commentary box talked about disruption. I took a stopwatch and timed every review. The average was 101 seconds. I cross-referenced fourteen VAR decisions across the tournament against the stoppage time of each match.

Average added time rose by only 2 minutes 37 seconds against the tournament baseline. VAR does not break the rhythm of a game the way people think. It concentrates dead time into one place, and most viewers never notice because that dead time coincides with the moment they are arguing.

The third study is the one I delayed longest. In June 2026, football returned after lockdown to empty stadiums. I collected data from 89 Premier League matches before and after the pandemic, comparing yellow cards and penalties per match.

Yellow cards fell 23 percent. Penalties rose 31 percent. With no crowd present, referees booked fewer players but pointed to the spot more often.

Reading those two figures correctly is the important part. Fewer yellow cards does not mean referees became lenient. More penalties does not mean referees became stricter. Both point the same way: without noise, referees judge what they actually see rather than what the crowd wants them to see. Yellow cards are largely a match-management tool used under pressure. Penalties are largely a pure decision about contact.

I also take notes live, never waiting for the replay. In a recent league match I counted nine advantages played instead of free kicks awarded, seven of which saw the fouled team lose possession within five seconds. The advantage rule is designed to protect the fouled team; in practice it often benefits the offender. That is the kind of systemic error that never shows up in the scoreline.

The same micro-method applies beyond refereeing. In November 2026, from the VIP stand during England's 6-2 win over Iran, I tracked Jude Bellingham while the stadium fixated on Bukayo Saka's hat-trick. Bellingham recorded 78 touches, 41 of them one-touch, never holding the ball longer than three seconds. I called a former scout to cross-check, then wrote a long-form piece on him before any major outlet picked up the thread.

CONTRARIAN ANGLE

The irony is that the data itself is now being misused in the opposite direction.

After every matchday, social accounts publish tables of VAR interventions, overturned decisions and added minutes, then conclude that technology is destroying football. Those tables are usually right on the numbers and wrong on the inference, because they measure the wrong quantity. They measure how often interventions happen. They do not measure how often interventions are correct.

There is a comparison I always use when asked about this. A referee gets one decision in forty-seven wrong. If VAR moves that to one in sixty, the referee is proportionally more accurate while more decisions are being reviewed. The only thing that changes is where the error lives, not how many errors exist. Mistakes do not disappear. They migrate from the pitch into the VAR room, where they are more expensive, slower, and replayed for millions.

That is the biggest blind spot in seven years of VAR debate. Both camps agree the goal is to reduce error. Neither admits football does not run on the logic of error reduction. It runs on the logic of risk allocation.

And here is where I turn the lens on myself. I was once a VAR sceptic, and that is why I understand the people who hate it. But looking back at the Anfield 2026 spreadsheet, I have to concede something uncomfortable: with VAR that day, Sunderland get a point, Liverpool lose two, and the match is worse for it. A correct decision can make a game poorer. I have no way to resolve that contradiction with data. I can only say I choose the correct decision, even when it costs me something.

TAKEAWAY

The direction of refereeing over the next three seasons will not be decided by technology. The cameras are good enough. The machines are fast enough. What is missing is consistency of interpretation, and consistency can only be built on public data.

My proposal is specific. Every match should publish a full referee ledger: total decisions, decisions reviewed, decisions overturned, and the reasoning. There is no need to release in-game audio, which belongs to the privacy of the person blowing the whistle. Publishing data belongs to the public, because the public pays to watch.

An empty stadium does not lose its soul; it returns the soul to its rightful owner. Data works the same way. When every decision is recorded and published, ownership of the match returns to the people who play it and the people who watch it, rather than the people standing in the middle.

The best referee is the one nobody mentions after the match. If data forces us to mention them more often, that is the price of finally knowing who is actually good.