The Evidence Threshold: What VAR Teaches Us About the Craft of Football Analysis

The Evidence Threshold: What VAR Teaches Us About the Craft of Football Analysis

In the 34th minute of the 2026 World Cup final at the Luzhniki Stadium, the score stood at 1-1. France won a corner on the left. The ball swung into the box, and when the replay rolled in slow motion, viewers saw it strike the hand of Ivan Perišić, the Croatia winger. Referee Néstor Pitana stood inside the penalty area, one hand pressed to his earpiece, eyes fixed on the monitor by the touchline. In the stands, more than 78,000 spectators roared. On the pitch, French and Croatian players waited in tense silence. Pitana had roughly sixty seconds to answer a single question: was this a penalty or not?

He gave the penalty. Antoine Griezmann stepped up, scored, France went 2-1 ahead and eventually lifted the trophy with a 4-2 win. But that is not the part I want to discuss. The part I want to discuss is the sixty seconds before the ball was placed on the spot, when Pitana had to choose between two irreversible rulings. That is the moment I always return to when someone asks how the craft of football analysis differs from the craft of fandom. The answer is not in the emotion of the stands. The answer is a principle: a referee must say "yes" or "no", and if there is not enough evidence to say "yes", the mandatory default is "no". I call that the evidence threshold. After 37 years watching this industry, after the 335 VAR interventions I recorded by hand in my own notebook, I believe the evidence threshold is the line between an analyst and a supporter.

In 2026, when a livestreaming platform invited me to serve as a rules expert for the World Cup in Russia, I built my own data table tracking all 64 matches. I logged every VAR intervention, every minute of stoppage time, every overturned decision. The final count: 335 interventions, 20 overturned decisions, 10 penalties originating from VAR. But the figure that troubled me most was the correction accuracy rate: 68.4 percent in the group stage, rising to 91.2 percent in the knockout rounds.

That gap is not about machines. It is about people. The group stage had 48 matches, lower pressure, teams unaccustomed to the tempo, and officiating crews still feeling their way. The knockout rounds had only 16 matches, where every decision could end a dream, and crews prepared far more carefully. What this means is that the quality of a conclusion does not depend on whether you have the tools, but on how thoroughly you have prepared your evidence threshold for that situation.

That is precisely the problem with football analysis today. We live in an age of overflowing data: expected goals, PPDA, passing maps, physical metrics. But abundant data does not mean correct conclusions. An analyst can cite twenty numbers in one paragraph and still be entirely wrong, because those numbers were not placed in context. The evidence threshold is quietly lowered, and the reader has no way to check the work.

I want to tell another story, this time not on the pitch but in the boardroom. In May 2026, when FIFA imposed a two-window transfer ban on Real Madrid and Atlético Madrid for breaching Article 19 of the Regulations on the Status and Transfer of Players, I used my economics background to estimate the damage. Real Madrid lost roughly 147 million euros in market opportunity. I wrote a 5,200-word piece with 47 article citations. Within 72 hours it reached 1.2 million reads.

What I learned from that article was not the 147 million figure. What I learned was the three-layer process: situation, statute, data. Every claim must carry a specific citation. If I have no statute, I have no article. If I have no data, I have no conclusion. And most importantly: if I have only emotion, I have nothing at all.

Three years later, when the COVID-19 pandemic halted global football for 97 days, I applied that same process to a harder problem. I built a "COVID-19 Force Majeure Tracking Table", surveying 386 player contracts across five major leagues and the Chinese Super League. When 38 contract-termination disputes were filed with FIFA's Dispute Resolution Chamber, I predicted only four would succeed. The actual result: five. I was off by one. But I did not treat that as failure, because I had stated my evidence threshold clearly. I did not promise certainty. I offered a testable prediction based on 386 contract samples and legal precedent. The difference between a prediction with an evidence threshold and a random guess is this: the first can be refuted with data, the second cannot.

This is what I want to stress as someone who works in legal commentary: the evidence threshold is not the analyst's weakness, but the measure of his professionalism.

Let us return to football. If I want to say a team has a fitness problem, I cannot simply say "they look tired". I must show that over the last three matches their PPDA rose from 8.4 to 12.1, meaning pressing intensity dropped by nearly a third. I must show that the gap between fixtures was only three days while their opponent rested five. I must show they played extra time in the domestic cup. Those three pieces of evidence combined create a conclusion strong enough to be printed in bold.

But if I have only one of those three, I must lower my conclusion threshold. I must write: "There are signs of declining intensity, but the sample is small and the opponent variable has not been ruled out." That is a hard sentence to write. It is not attractive. It will not generate a headline. But it is honest.

The problem with football analysis today is that very few people dare to write that sentence. Algorithmic pressure rewards decisive headlines. "This team will certainly be relegated." "That player is finished." Those lines get shared ten times more than a conditional one. And readers, after years of exposure to such decisive lines, slowly forget that most sporting conclusions hold only within a certain probability range.

That is why I always keep raw data intact in my writing. I do not round. I do not trim the sample. I leave the 386 contracts, the 335 VAR interventions, the 68.4 percent and the 91.2 percent exactly where they are in the text. Readers have the right to verify for themselves, as if scanning every millimetre of VAR footage. If I hide the sample, I have stolen their right to check.

There is an objection I often hear: if an analyst keeps saying "not enough data", what is left to read? Readers want answers, not hesitation.

I understand that objection, but I believe it confuses two things. The hesitation of a lazy person is entirely different from the caution of a professional. The lazy person says "not enough data" to avoid work. The professional says "not enough data" after checking every source, and simultaneously specifies what further data is needed to reach a conclusion. A properly stated "not enough data" must come with a list: the competition name, the current standings, the last three matches, the publication date.

Referees are the same. When Pitana lacked enough evidence to award a penalty, he did not silently wave it away. He went to the touchline monitor, reviewed the footage, and if it remained unclear, he upheld his original decision. He did not guess. Nor did he refuse responsibility. He followed the process.

What fans in the stands do not see is that behind a "no penalty" decision there may be thirty seconds of review from four different camera angles. Viewers only see the outcome. A good analyst must show the reader the process, not just the result.

And this is where emotion and rule split the road. Emotion says: this team deserves it. Rule says: there is no evidence. Between those two voices, an honest analyst must choose the second, even knowing it will disappoint half his readers.

I once witnessed this in its most extreme form in 2026, when three Chinese clubs called me for urgent advice on their pandemic-era contract disputes. They wanted me to say they would win. I said that according to 386 data samples, their chances of winning were low. That was not the answer they wanted. But within three days I compiled a 17-page crisis-management standard, listing five scenarios, five statutes and five concrete action plans. Several clubs followed it and avoided heavier sanctions.

The lesson here is institutional. A mature football ecosystem needs more than good referees on the pitch. It needs good analysts in the boardroom, people willing to say "I do not have enough data" when that is the truth. And it needs audiences who accept that a conditional answer can still be an honest one.

After 37 years watching football from Belgrade to Shanghai, I have drawn one conclusion: football analysis will not advance by producing more conclusions, but by raising the evidence threshold for each one. A football ecosystem may nurture thousands of analysts, but only a few dare to write the sentence "I do not have enough data to conclude".

If you read a sports analysis today and cannot find a single verifiable number, ask yourself: is this analysis, or just noise dressed up in words? And if you write for a living as I do, ask the harder question: did my most recent article clear its own evidence threshold? Because through the referee's eye, you cheer for no one. You only look for who is right.