From Liverpool's 91 goals to Croatia 2026: the cost of empty analysis

On April 24, 2026, at Anfield, Liverpool beat Roma 5-2 in the first leg of the Champions League semi-final. Five goals, five names: Mohamed Salah twice, Roberto Firmino twice, Sadio Mané once. That night, a line I had written on my personal blog in August 2026 — that this front three would score at least 84 goals in all competitions — suddenly had a date and an address. By the end of the 2026-18 season, the final figure was 91: Salah 44, Firmino 27, Mané 20. People call me reckless, but the numbers have never lied. I retell that old story to talk about a habit spreading fast through the football content trade. The conclusion gets fired off first, the data is hunted afterwards, and when no data can be found, the conclusion simply stays where it is. Nobody comes back to check. Readers finish it, forget it, and move to the next one. Based on my experience watching matches in England and notebooks that begin in 2026, that loop has only three steps. A match ends. A verdict appears within ten minutes. A week later nobody remembers whether it was right, only how sharp it sounded. In Vietnam the loop runs even faster, because English-language sources reach the writer almost instantly. What worries me is not the volume of opinions. Football lives on argument. The problem appears when conclusions detach completely from evidence, analysis turns into guesswork, and readers lose the ability to tell who actually knows what they are talking about. The three cases below are three different ways a verdict either holds up or collapses, depending on whether it carries numbers. Case one: one hundred million pounds and 91 goals. In June 2026, Liverpool paid Roma 36.9 million pounds for Salah. In June 2026, Mané arrived from Southampton for 34 million pounds. In June 2026, Firmino arrived from Hoffenheim for 29 million pounds. Roughly 100 million pounds for three men. In 2026-18 they returned 91 goals across four competitions — under 1.1 million pounds per goal. In the summer 2026 transfer market, when a full-back could cost 50 million pounds, that division looked absurd. But if I stop at the division, I commit exactly the error I am criticising. A hundred million pounds for three players only produces 91 goals when they are placed inside a specific system. Jürgen Klopp built that attack around Firmino's false nine role: the Brazilian dropped deep to drag centre-backs out of position, opening space for Salah to cut in from the right on his left foot and Mané to surge in from the left. Without Firmino dropping, Salah has no land to run into. Without Salah and Mané making those runs, Firmino is a harmless deep-lying forward. Numbers do not explain themselves. They only record the output of a machine somebody assembled. Case two: Croatia and the luck label. At the 2026 World Cup, Croatia went through three knockout rounds and all three went to extra time: 1-1 with Denmark in the round of 16, 2-2 with Russia in the quarter-final, 2-1 with England in the semi-final. The popular version afterwards was that Croatia were lucky, drew an easy bracket, and won on penalties and character. Passing data from the knockout rounds for the midfield trio Luka Modrić, Ivan Rakitić and Marcelo Brozović shows an accuracy rate of about 89 percent. That is why Croatia dragged hard matches into extra time and stayed there longer than their opponents. Keeping the ball is not luck. It is a measurable tactical choice. I once got the 2026 World Cup wrong — I mispronounced Rakitić's name three times in a row on live commentary, then spent the following month rewatching footage and recounting every Croatian midfield pass. My error then was not the pronunciation. It was that I spoke about a team for which I had not prepared enough data to speak. Case three: when the stands were empty. For most of the 2026-21 season, Premier League stadiums were closed to spectators. Home win rates fell to their lowest level in decades, while away wins rose correspondingly. Before that, the league's long-term average sat around 45 percent of matches ending with the home side winning. Half of that season's table was rewritten by a variable sitting outside the pitch. Prediction models that used only on-pitch data got match after match wrong. Home advantage is dead — at least the version we thought we understood. Key point: A conclusion only has value when it is tied to something verifiable — a statistic, a date, or an observation specific enough to be wrong. That is the whole argument of this piece, and it sounds so obvious that people will nod and move on. Anyone can be wrong, including the man holding the scales. Data looks backwards. It counts what happened; it cannot measure the moment a centre-back loses composure after the 80th minute, the noise that makes an opposing goalkeeper misplace a pass, or a team playing upward out of fear of losing. A stadium holds a layer of information that a spreadsheet never touches. And my own 84-goal prediction from 2026 deserves a fair audit too. It was written from watching Liverpool play, not from any regression model. I saw something in them before the world turned its head, then translated that sight into a precise number. Data does not kill emotion. It gives emotion a frame. The remaining trap sits on the opposite side. When writers become obsessed with always having a figure, they slide into one of two ruts: stuffing numbers where none are needed, or meeting a data-poor topic and choosing to write a fake conclusion with a few decorative stats on top. The most honest route, and the hardest to sell, is to say plainly that there is not enough information to conclude. An analyst who says that is still more useful than one who invents a tidy answer. Football waits for no one. It only waits for those willing to ask questions. Next season I will do something small and checkable: count how many verdicts on major English broadcasts and bulletins come with at least one specific, sourced figure attached. If that share is under one third, I will say plainly that football's analytical ceiling is far lower than its surface suggests. And I will publish the result, even if it goes against me.