The Conclusion Addiction: Football Analysis Talks Loudly About What It Never Knew

In the 67th minute, the big screen at Groupama Stadium flashed a line of data: 68% possession for the home side. The crowd applauded as if a goal had been scored. I was sitting in row seven of the press area, looking down at the pitch, and I watched the home side's central midfielder complete his thirty-fourth sideways pass of the second half, forty-two metres from the opponent's goal, with no red shirt within an eight-metre radius. A perfect pass, aimed at an idea that did not exist. Three minutes later the away side broke and scored their second. The possession figure on the screen still looked as handsome as a life-insurance billboard. The colleague next to me typed the opening line of his post-match column: "The home side controlled the game but lacked efficiency." He had just finished building a complete analytical frame around a match he had not actually watched. A Ligue 1 fixture ends at 22:55. By 23:10 the post-match piece is live. By 23:30 three pundits are seated at the desk. By midnight the verdict has been packaged, tagged and shared. None of them had time to rewatch the first half. Demand for conclusions exceeds the supply of evidence, and when demand outstrips supply, the market does the only thing it knows: it sells the shell. A formation graphic, three bullet points, two mentions of "high pressing", and there is a product that looks like analysis. Underneath the paint there is usually nothing. I call it the empty bulletin. In 2026 I started out at a local radio station, with a cassette recorder and a phone book. We had no data. We had eyes, we had phone calls at eleven at night, and we had a thing now called professional shame at speaking about something we had not seen with our own eyes. Thirty-two years later the industry has Opta, Stats Perform, hundreds of metrics, and a new habit: build the frame first, insert the evidence afterwards. If the evidence never arrives, the frame still stands, still confident, still shareable. In France they call it the débrief. In Vietnam they call it a chat-room livestream. The two football cultures differ in almost everything: budgets, climate, wages, even the way crowds shout a player's name from the stands. They share one fatal trait. Both are addicted to conclusions. A V.League match ends and fifteen minutes later twelve threads of opinion have appeared online, at least nine of them written by someone who only watched the goals on a phone screen. A Ligue 1 match is the same, except the writers wear better coats and hold a press licence. Readers do not check. That is the entire operating mechanism of the empty bulletin. A tidy argument frame reads exactly like real analysis, because its shape is correct. The error lives in its guts. On 1 July 2026, Spain met Russia in the World Cup round of sixteen at the Luzhniki Stadium, in front of more than seventy-eight thousand spectators. Spain completed over a thousand passes in that match. Russia completed fewer than three hundred. Possession leaned towards Spain at a level television usually calls domination. After 120 minutes the score was 1-1. Russia won the shootout 4-3, goalkeeper Igor Akinfeev saving two attempts. I watched that game from start to finish, and what I remember is not the pass count but the sound: the sigh of the stands every time the ball went back to a centre-back. Those passes were all counted, all accumulated, all printed on the stats board as proof of dominance. In reality they were proof that nobody dared to open a road. Possession measures exactly one thing: which team touched the ball more. It does not measure who came closer to scoring, who controlled the dangerous zones, who forced the opponent to change their plan after twenty minutes. A team grinding out 60% with sideways passes in its own half owns a beautiful metric and a large problem. The crowd believes in the numbers board. I believe in the pain on the grass, in the moment a defender turns to look for a teammate and finds nobody there. Then there is xG, treated as gospel for the past decade or so. Expected goals models are built from hundreds of thousands of past shots, assigning each attempt a scoring probability based on location, angle, body part and the type of pass that preceded it. Statistically it is rigorous. Humanly, it assumes an average footballer who does not exist. The same shot from the same position at the same angle, taken by a thirty-three-year-old defender who has just missed two, and by a striker who has scored forty this season, receives the same xG value. A decade ago I believed in data. Now I believe in my eyes. Not because the data is wrong, but because the data is being read as if it knows more than it was ever programmed to know. There is one more thing few people want to say: xG depends on match state. A team leading 2-0 takes fewer risks, shoots less, and the figure drops. A team losing 0-2 pushes forward, shoots more, and the figure rises. At full time, that figure is used to argue the losing side deserved more. A tool that appears independent of the scoreline has quietly read the scoreline first and retold it in scientific language. I once sat in a studio where the graphics operator pre-loaded the "high pressing" template before kick-off, because he knew it would be needed regardless. It was a completely rational action inside a completely broken system. The frame is built first, the content is searched for afterwards. When nothing is found, they broadcast anyway, because the frame is already in the machine. The empty bulletin does not only live on television. It thrives best in the transfer market. In August 2026 Paris Saint-Germain paid 222 million euros for Neymar, breaking the world record. At the same time, in Monaco, an eighteen-year-old named Kylian Mbappé had just scored 15 Ligue 1 goals in his first season as a genuine starter. I wrote on my personal blog that he was worth more than the 222-million-euro item. I received hundreds of jeers from accounts calling me a madman. A few weeks later PSG completed the Mbappé deal for 180 million euros. The transfer market does not run on mathematics. It runs on psychology with a spreadsheet attached. A transfer is not an addition problem, it is a negotiation between two fears. The Saudi Pro League is the cleanest example of using money to buy a story. Since Cristiano Ronaldo signed for Al Nassr in January 2026, the league has imported a generation of stars past their peak. Those contracts do not exist to develop a football culture, they exist to create ambassadors with legs. A league cannot buy an identity with two-year contracts, just as a news bulletin cannot buy the truth by printing its numbers in bold. The analytics industry is now walking into the dressing room, carrying reports prepared on Monday for a match on Saturday. In the gap between those two days, a centre-back can pull a hamstring, a midfielder can lose sleep over a sick child, a young player can just have learned his wife is pregnant. No model enters those things into the spreadsheet. Based on my experience watching matches, I see this most clearly at clubs rated as the best prepared: they control every variable off the pitch and are powerless against the variable inside a player's chest. People call me a contrarian. I call them the crowd. But I have been wrong, and wrong in the most memorable way in May 2026. In March of that year, when the leagues stopped because of the pandemic, I wrote that football without spectators was a fake product and called for the season to be cancelled. I wrote a line I still like: a stadium without a crowd is just a car park painted green. In May the Bundesliga returned. I watched the Ruhr derby between Dortmund and Schalke inside an echoing empty ground. The players threw themselves at each other as if eighty thousand people were screaming their names. I publicly retracted my position on my own blog. A few readers unfollowed. A few colleagues began to look at me differently. I learned something the me of ten years earlier would never have believed: the most dangerous thing in this trade is not being wrong, it is being wrong and never having to pay for it. Data has also caught things my eyes missed. Liverpool appointed Thomas Grønnemark as a dedicated throw-in coach in 2026, a decision the English press treated as a joke at the time, and it produced measurable gains. Brentford built an entire system on set-piece analysis to climb up from the lower divisions. My traditional eye has blind spots too, and admitting that does not make me smaller. So my position is not that data is useless. I am saying data is being sold as a religion when it is only a tool. And conversely, "I trust my eyes" is also an argument frame, the laziest one of all, if it never has to answer a single metric. When the whole world speaks in unison, my ears start ringing with the echo of error. But when I am the only one speaking, I also have to ask whether the ringing means I am right, or means I am deaf. There is one small detail about myself I should tell, because it belongs to this subject. On the night of the 2026 World Cup final in Lusail, I wrote "ninety thousand people" in the dispatch I filed from the stadium. The actual capacity was 88,966. I was off by more than a thousand, a small error in a long piece, but it came from exactly the habit I am criticising: trusting an impression rather than trusting the page. A contrarian can also be lazy about checking. I know that about myself, and I choose to live with it, because being caught out always teaches more than being praised. I do not write to be loved. I write to be read. I want to offer a verifiable test for the next World Cup, rather than a vague promise. Within forty-eight hours of a knockout match, at least one metric will be cited by most outlets as the decisive cause, and then contradicted by deeper data from the same provider. A more specific prediction: a team with under 45% possession will eliminate a title favourite, the media will call it character, while chance-quality data shows the two sides were nearly level. Anyone who wants to check on me can open the book and write it down. I write to be read, but I also write so that someone opens the book and checks.