Four Rounds of Premier League and the Trap of Early Verdicts

September, a stadium in the north of England. Fine rain, the smell of cheap beer mixing with wet grass. The home side had just lost their third straight match at home, and the man beside me, striped shirt, wool hat, about sixty, screamed something at the row in front that I could not translate because the local accent was too thick. But I understood the meaning. He was sacking the manager, sacking the board, sacking his club's entire twentieth century. All of it inside four matchweeks. I pulled out my phone. The table loaded. Three points, one point, none, three. And on the screen, a commentary headline was calling this club a team in trouble. Four matches. Three hundred and sixty minutes of football, minus stoppage time. People call me a contrarian. I call them the crowd. Every season is the same. Around matchweek four to six, a familiar kind of content surfaces across sports newsrooms: who has had the best start, who is in trouble. The template barely changes. A few lines of setup, a few selective numbers, then an invitation for readers to comment. A line like you know the drill by now appears, and the crowd piles in. I work in France, writing for the French market, but I track how English outlets operate because the Premier League long ago became a global television product more than a football competition. That engagement format has a very clear function, and the function has nothing to do with football: it measures the emotional temperature of the audience, harvests comments, and then uses those comments as raw material for the next piece. Fans comment, the outlet aggregates, fans comment on the aggregation. A closed loop with no room for data. I know this game from the inside. In 2026, aged forty-one, I wrote a deliberately provocative blog post: eighteen-year-old Mbappé at Monaco was worth more than Neymar at two hundred and twenty-two million euros. At the time Mbappé had scored fifteen Ligue 1 goals, and Neymar had just broken the world transfer record. Hundreds of accounts called me insane. Weeks later, Paris Saint-Germain paid one hundred and eighty million euros for Mbappé. Lesson one: a shocking statistic plus a famous name generates enormous debate. Lesson two, which few noticed: enormous debate does not equal correct analysis. I was right because I read the data and read the player. I became famous because of the headline. Two different things, and the crowd only remembers the second. Technically, the best start versus in trouble framing after four rounds is a probability game dressed up as a moral verdict. I am not talking about emotion. I am talking about numbers. A Premier League season has thirty-eight matchweeks. Four rounds is just over ten percent. In statistics, ten percent of a sequence is not enough to separate signal from noise. Four matches carry too many variables: a wrong penalty, a tenth-minute red card, a goalkeeper having a bad day, a player missing a flight because an airport closed. Add it all up, and the variance is high enough that a mid-table side can sit third after four rounds and finish twelfth without ever playing worse. Conversely, a big club can sit fifteenth in early September because it faced three of the top four in its first four fixtures. This is where advanced data should appear, and almost never does in that emotional commentary format. Expected goals measures chance quality rather than outcome. A team scoring three from 1.2 xG is lucky. A team scoring one from 2.4 xG is playing well. After four rounds, the gap between actual points and expected points is usually far wider than the gap between fourth place and fourteenth place at season's end. In other words, the early-September table says far less about the teams than about the fixture list. And the fixture list is the thing that format almost always ignores. A team labelled in trouble may have played three Champions League sides in four games. A team labelled best start may have faced four promoted or struggling clubs. Normalise for opponent difficulty and the story often flips. I have run this comparison many times in my own notes, and the reversal rate is high enough that I have stopped being surprised. A decade ago I believed in data. Now I believe my eyes. That sounds like a rejection of statistics, but it is not. I trust my eyes because eyes see what the table hides, and the table sees what eyes hide. What I lost faith in is not data. What I lost faith in is the way people stuff data into a pre-written story. I watch early-season matches by rewatching the full ninety, not just the goals. It is brutally time-consuming and I advise nobody to copy it. But it taught me something the table cannot: a team can win three of four while its pressing structure has already collapsed. A team can lose three of four while its transition system has clearly taken shape. The first will be figured out by November. The second will be fine. No table tells you that, because a table counts outcomes, not processes. There is one metric I prefer above all: the number of attacking sequences a team generates after winning the ball in the opponent's half. It measures the ability to convert chaos into chances. After four rounds it is noise. After ten it starts to speak. After twenty it is almost always right. The boundary between noise and signal is a measurable threshold, not an opinion. I use that threshold as my standard. Eight to ten matchweeks. That is the minimum sample before calling a start a trend. Before that mark, every verdict is a guess dressed in a confident voice. And a confident voice is the cheapest thing in this trade. You do not need to go to the ground. You do not need to watch the match. You just open the table, pick two clubs at opposite ends, and write. That explains why the format exists. It does not exist because it is right. It exists because it is cheap. A reporter in an office can produce it in forty minutes, and it generates more engagement than an xG analysis that takes two days. The economics of sports journalism reward speed and emotion, not accuracy. I am not making a moral judgement. I am saying readers should know what they are consuming. In France, where I live, people play the same game with a different accent. French outlets prefer the crisis frame over the best-start frame. England prefers the fairytale. Both come off the same assembly line: turning a small sample into destiny. I grew up in Vietnam, where fans talk about form and fate as if they were physical objects. In France, people talk about spirit and character. Same belief, different language. And all of it is how humans try to find meaning in three hundred and sixty random minutes. When the whole world speaks in unison, my ears start ringing with the echo of error. Now the part where I might be wrong. There is one case where a small sample genuinely predicted a large one, and I have to acknowledge it because I refuse to be a contrarian out of ego. Leicester, 2026-2026. After ten rounds, nobody believed. By season's end, they were champions. But note the threshold: ten rounds, not four. And note how they won, through a ferociously disciplined defensive-transition system. That is my point: a small sample predicts when it contains a structure, not when it contains a result. I can be wrong elsewhere too. Sometimes the who-is-in-trouble format catches a real problem that data has not yet priced in: a dressing-room crisis, an owner who has stopped spending, a manager losing control. Data arrives late. The fan's eye arrives early. I have been wrong in exactly this way. In 2026 I wrote that football without fans was a cheap imitation and called for the season to be cancelled. In May, the Bundesliga returned to empty stadiums. I watched the Ruhr derby between Dortmund and Schalke, not a soul in the stands, and the players still threw themselves at each other as if eighty thousand were roaring. I was wrong. I publicly retracted, lost some readers, and learned that lived experience sometimes outruns your own momentary feeling. So when I say the September table is a media product, I am not saying every early observation is worthless. I am saying most of them are the output of a format, and that format was never designed to be right. My testable prediction: of the clubs in the best-start group at the four-to-six matchweek mark, at least half will have dropped out of that position before matchweek twenty. And of the clubs labelled in trouble, at least half will finish the season in the top half of the table. If that does not happen, I will write a retraction, as I did in 2026. I do not write to be loved, I write to be read. But I want you to read with open eyes, not with a comment already typed.