Right Framework, Empty Evidence: Why Football Analysis Must Begin With Data

On my third rewatch of the same match, I kept the notebook closed and added nothing. Three viewings, three times the same question: was that gap the coach's intention, or just the result of eleven tired men running out of sync? I could not answer. And instead of inventing a plausible-sounding answer, I closed the notebook. That is the whole job: knowing when you do not yet have enough evidence to write. There is a kind of document I encounter more and more often. It is beautiful. It has headers, columns, tables, nine major sections and dozens of smaller rows. It obeys every formal standard a professional analysis is supposed to obey. But when you read the most important line of all — the line that should hold the facts — you find a single short word: empty. No team name, no player name, no minute, no score, no number to hold on to. The frame is perfect. The inside is hollow. I remember the first time I saw one. It made me strangely calm, and then genuinely worried. Because an empty analysis harms no one while it sits alone. It only harms people when a reader sees the frame, trusts it, and assumes the inside deserves the same trust as the outside. That is the moment a document that says nothing begins to be treated as a document that has said everything. In football, we have another name for that moment. We call it a smooth night of analysis. The context here is not a single match but an entire ecosystem. Twenty years ago, a viewer had two sources: the broadcast and the commentator. If you wanted to know why a team won, you had to trust the person talking. Today it is inverted. Data pours out everywhere: passes, heat maps, line distances, expected goals, pressing actions per minute. A person sitting in Hanoi can rewatch a Paris team at three in the morning, pause at minute 63, rewind three seconds, and count. That is a revolution. But every revolution carries a debt. The debt here is this: when data becomes easy, people start believing data itself is analysis. It is not. Data is raw material. Analysis is the process of turning raw material into a judgment that can be disproved. Between those two things lies a gap far wider than it looks, and that gap is exactly where hollow frames are born. I think of something small that haunted me for years. When I started writing, I once built a piece on a feeling about a match. It read smoothly. Every sentence sounded right. Readers praised it. Then I rewound the match and discovered that one of my key observations — the prettiest one, the line I loved most — did not actually happen at the minute I claimed. It happened at another minute, in another situation, with another player standing where I thought someone else stood. I was right about the feeling and wrong about the fact. Since that day I set a rule for myself: every number must be checked three times, against two independent video sources, before it is allowed into a piece. That rule did not make me write slower. It made me write differently. It forced me to filter, to drop observations I could not verify, to accept that a piece with fewer ideas but where each idea stands firm is better than a piece with many ideas that cannot survive one simple question: where did you get that number? There is one event I always return to when I need to teach myself data discipline. France against Argentina in the round of sixteen of the 2026 World Cup in Russia, on June 30, 2026. I was seventeen, sitting in front of a screen with a squared notebook. I wrote down the final score: France won 4-3. But what stopped me was not the score. It was possession: France held the ball only about 39 percent of the time. A team that wins without the ball. A strong team that yields the pitch. That night I began doing something that became a habit: I stopped watching the ball and watched the players without it. I recorded the positions of the French players while the opponent had the ball. I redrew the shape after the match. What I saw were the spaces behind Argentina's two full-backs — spaces that did not appear by accident, but as the result of a decision: France agreed to sit deep to lure Argentina forward, then used speed to attack the very space Argentina had just created by pushing up. The tactical map I drew was born on a France–Argentina night, where two shirt colors merged into a single intention. From that night, I understood that a match is not read by counting goals, but by tracing the intent behind every meter of ground left empty. But the most important thing I learned that night was the opposite of excitement. It was the cold feeling of asking myself: does the intention I just drew actually exist in the coach's head, or am I seeing my own hypothesis reflected on the screen? For a decision to be called a decision, it must repeat. Not once, in one beautiful sequence. But as a pattern of behavior, repeating, measurable, predictable before it happens. That is the line between an analyst and a storyteller. The storyteller sees a gap and tells a story about it. The analyst sees a gap, then looks for at least a second occurrence and a supporting data point before allowing himself to name it. In March 2026, European football stopped. I was nineteen, living in Paris, and suddenly had a long stretch with no new match to watch. I decided to use that time for something nobody normally has time for: rewatching an entire season of one team. Four months frozen, I sat with PSG fifty-seven times to hear them speak through gaps. I did not watch to find beautiful moments. I watched to find habits. I divided the pitch into twelve zones and recorded each player's pressing frequency in each zone. I counted how often an Italian midfielder — Marco Verratti — pushed forward to press behind an opposing midfielder, and how often he chose to stand still. I logged successful presses, presses that left gaps, and presses aimed at the right man but at the wrong moment. After finishing, I wrote a piece titled around PSG's pressing trap. It was published on a new sports platform and drew more than ten thousand views. But the number I remember most is not those ten thousand views. It is the three revisions I had to make before publishing, only because I found measurement errors in the distance between lines. Three times. For a piece people read in five minutes, I spent nearly a day just fixing the meters. Some people told me that was a waste of time. I think the opposite. Precisely because people read it in five minutes, I am not allowed to be wrong. If I get a number wrong, the reader has no way to detect it, and because they cannot detect it, that error will outlive me. A piece that is wrong about specialist data is a debt left to the reader afterward. From that I drew a principle I still keep today: if a number cannot be verified by two independent sources, it has no right to enter the piece. Not because I doubt everything, but because I respect the reader enough not to ask them to believe what I have not verified. In 2026, at the World Cup in Qatar, I had a chance to test that entire set of principles against the most complex subject: a team the whole world called negative in defense, while in reality they were playing an extremely proactive defensive system. That was Morocco. The North African side reached the semifinal and on the way conceded exactly one goal — an own goal against Canada. Coach Walid Regragui built a 4-1-4-1 with captain and center-back Romain Saïss, thirty-four years old, commanding the back line. The world's media called it a wall. I wanted to know how thick that wall was and how it moved. Morocco built a wall, and I was the one writing a diary for each brick. I measured the average distance between Morocco's lines. About twenty-eight meters. A small number. But more important than the number was whether it stood still. It did not. That mobile wall slid laterally with the ball, narrowing when the opponent held the ball centrally, widening when the ball went wide, always keeping a line of bodies between the ball and the goal. I wrote a piece pushing back on the term negative defense. I argued that Morocco defended with intent, counter-attacked with intent, and that the ugliness people attached to them was a prejudice about aesthetics rather than a judgment about tactics. They did not defend because they could not attack. They defended because they chose the moment to attack. A veteran expert shared the piece. Someone called wanting an interview. I declined. Not out of arrogance. Because I wanted time to recheck all my measurements once more before saying anything else. Here, my story meets the story of that hollow document from the start. In recent years, professional football analysis has industrialized so fast that it has produced a whole standard set of frames. That is good, in a very specific sense. With frames, practitioners do not have to rebuild from scratch each time. A properly standardized analysis will have a tactics section, a finance section, a results-and-reaction section, a league-landscape section, a rules-and-governance section, a management-and-dressing-room section, a risk section, a media section, an industry-transmission section. Nine sections. Each with boxes, columns, rows to fill. The problem is born right there. When a frame becomes complete enough to look like a product by itself, pressure appears to fill it with anything, including sentences that carry no facts. A table with full headers, full rows, full marks, but every cell empty, still looks more professional than a short paragraph saying plainly that we do not yet know anything. That is the first blind spot I want to put on the table. A frame is not evidence. A frame is only a stand on which to hang evidence. An analysis with nine sections but no facts is not an unfinished analysis. It is an analysis that has not begun, presented in a way that makes people believe it has ended. And when such a document enters a larger system — a digest, a tracking sheet, a feed that flows into later decisions — the risk of propagation begins. I call it data contamination. An empty document, if labeled complete, becomes the foundation for the next one. The error is not that it is wrong. The error is that it looks right. In twelve years in this work — including the years when I was only taking notes alone — I learned that the most dangerous error in analysis is not an error of conclusion. The most dangerous error is an error of foundation. A wrong conclusion can be disproved by the next match. An empty foundation cannot be disproved, because it says nothing. It simply waits for someone to build on it, and everything built on it carries that emptiness. There is one word I use constantly and see constantly misunderstood. The word is: highlight. Many think a highlight is the best clip, the prettiest goal, a moment to post for fun. To me, a highlight is a sample of evidence. It is a specific event, with a minute, a position, a person, a ball, saved so that a habit can be traced. When I suspect a team is pressing the wrong line, I do not argue with feeling. I find a highlight of that wrong press, then a second highlight of the same error. Two identical highlights are a hypothesis. Three identical highlights are a principle. And a principle is only then worth writing about. That is why I am never satisfied with a result. The score is the surface layer. I want to go below it. I want to know who decided what, at which minute, and why they chose that instead of the opposite. But precisely because I go that deep, the line between a correct hypothesis and a beautiful illusion becomes fragile. This is the trap people who work like me fall into most: reading gaps as intent. We believe the match speaks through gaps, so we easily assign each gap an intentional tactical purpose. But not every gap is intent. Some gaps are just fatigue. Some are just miscommunication. Some are errors, and errors are not tactics. So I set a disproof condition for myself. If I claim a gap is intent, I must find at least two repetitions, or a supporting data point from a source that is not my own eyes. If I cannot, that sentence is struck from the piece. Not because it is wrong, but because it is not yet qualified to exist. This sounds simple. But it transforms how an analysis gets written. Because once you force every claim to carry a disproof condition, you stop writing lines that sound good but cannot be defended. You start writing less, slower, more precisely. And you start treating uncertainty as part of the product rather than a weakness to hide. In France, where I live and work, I see two schools of analysis coexisting. The first is the emotional school: tell the match as a story, find a hero, find a tragedy, end with a beautiful line. This school has enormous pull, and I understand why. Football is emotion. But when emotion becomes method, it quickly slides into personal praise, and I do not want to write that kind of piece. I care about structure and on-pitch decisions more than the beauty of a player. The second is the frame-and-data school: sort everything into boxes, measure everything with numbers, verify everything with metrics. This school has verification power, but it carries a trap opposite to the emotional one. That trap is data overload destroying the narrative. The reader drowns in numbers and no longer remembers how the match actually unfolded. I try to walk between them. Every piece needs an anchor: one situation, one character, one specific minute, to keep the whole thing from drifting into a sea of statistics. But every claim must rest on data. The anchor keeps the narrative. The data keeps the truth. They are not rivals; they complete each other. Across fifty-seven PSG matches, the only thing they never rewatched was their own fear. I write that line not to show off. I write it because it is what I genuinely saw on rewatch. A strong team, at decisive moments, often shrinks before the opponent can even apply pressure. Fear appears before the danger does. And on film, that fear shows up in no number on a stats sheet. It shows up as a gap on the pitch: a midfielder who should push forward stands still, a defender who should hold the line drops half a meter. Precisely because it does not show in statistics, it is the easiest thing to overlook, and also the easiest to over-paint if the analyst wants a beautiful story. This is where I must remind myself to stop. Once I have called a gap fear, I am entering psychology, where I have no data. I can observe behavior, but I cannot read minds. So I choose to write cautiously without blurring. I clearly separate two kinds of sentences: claims and hypotheses. A claim needs data behind it. A hypothesis must be marked as a hypothesis, with the condition under which it would fail. When I do that, I no longer have to choose between certainty and honesty. There is another limit I always keep: never turn a piece into personal praise. This is especially hard in a major tournament, when collective emotion rises and everyone wants a hero. Football is a team sport, and one outstanding player cannot save a wrong structure. If I praise a player, I must show which structure let him shine. If I criticize a player, I must show which structure pushed him into trouble. No individual exists apart from the system. This is also why I keep a professional distance from what I call underlying positions. I have my own views on many things in the industry: on pressing that has been decoded to the point of becoming athletics, on representation contracts that stop players from saying what they think, on academies opened by former stars that are mostly commercial stunts while systematic grassroots coach education is badly underfunded. But I do not turn those views into slogans. I let them surface through which topics I choose, which data I pick, which match I dissect. A good piece is not the loudest one. It is the one that leads the reader to a conclusion without the author declaring it. Now I can say plainly what that hollow document made me think about. Across modern football analysis, there is a product type I consider more dangerous than a wrong prediction. It is a product that predicts nothing, yet is presented as if it has predicted enough. It has a conclusion, but the conclusion attaches to no evidence. It has a risk section, but risks are listed without level, likelihood, or impact. It has a data section, but the data is one word: empty. I once saw such a document in a professional discussion. It was about a big match. Many highly competent people sat around. And I stayed quiet until near the end. That is my habit: rarely speaking early, but when I do, often going against the room's consensus. The room's consensus was a beautiful conclusion. Very reasonable-sounding, delivered in a very confident voice. Nobody asked what it rested on. I asked. I asked one simple question: how many times did we see this in the match, and at which minutes? The answer was a silence. Not because the person did not know football. But because the conclusion was built from a general feeling, not from repeated observation. And all of us realized it at the same moment. That silence was worth more than the conclusion. It was the moment a hollow frame revealed itself as a hollow frame. This is the counterintuitive point I want to keep for the end. People usually think an analyst's value lies in original conclusions, in saying what others did not think of. I think the opposite. The greatest value of an analyst lies in the ability to say two words: not enough. Not enough data to conclude. Not enough sample to generalize. Not enough verification to speak. The ability to say not enough is the hardest to train, because it goes against the pressure of the profession. This trade rewards those who dare to speak. Shows need commentary. Sites need output. Nobody pays for a silent evening. But if you want to keep your credibility over years, you must learn to be silent at the right moment. Looking back over a long road, I see my career split into three kinds of pieces. The first is the feeling piece. When young, I wrote many. They read easily, they had rhythm, but they could not withstand the pressure of a data question. The second is the data piece. More solid, but with data alone and no story, it is dry as minutes. The third is the data piece told through a situational anchor. That is what I try to write now. And the third kind needs a mediator I call narrative discipline. That discipline includes a few very concrete things. One: each piece must answer a single question. Not three. One. If I cannot define it before writing, the piece will thin out. Two: each claim must come with at least one concrete fact: minute, position, name, number. Without a fact, the claim must be isolated as a hypothesis. Three: each piece must state the condition under which its conclusion would be overturned. If I say a team defends by lines, I must also say that if in the next match they switch to man-marking, the hypothesis fails. Four: each piece must have an emotional anchor. Not my emotion, but the match's: a tense minute, a human moment, a risky decision. The anchor makes the piece memorable, but it must not replace data. Five: each piece must close with a question pointing forward, not a summary. Because football does not end with the last match. It continues in the next one, where every hypothesis gets a chance to be tested. That is the whole spirit of the method I apply. At a wider scale, I see this shift as more than my personal story. It is the story of a profession having to redefine itself as data and automated tools rise. When machines can produce fluent text in seconds, human value stops lying in producing words. It lies in guaranteeing that every word stands on a fact that can be questioned. The competitive skill of the human analyst will no longer be writing beautifully. It will be verification. Knowing what is evidence, what is hypothesis, what is decoration. Decoration, anyone can do. Evidence must be earned. There is something I always think about my Vietnam, even though I live in France. In a football culture with limited resources, the scarcest thing is not passion. Passion abounds. The scarcest thing is data discipline. We have many people who love football and talk about it beautifully, but we still lack people who will sit with one match for hours just to make sure the number they are about to write is correct. And I think that is exactly what a developing football culture needs to popularize, like popularizing literacy. Because if I had to pick one lesson from everything I have lived, from a France–Argentina night at seventeen to Morocco's mobile wall at twenty-one, it is this: a conclusion without a foundation is not frightening, it is only pitiable. What is frightening is a beautiful document. Because a beautiful document spreads, and the emptiness inside it spreads with it. Transfers are where people buy players, while coaching staffs buy time. I add one thought to that: the analyst buys trust, and the only currency to pay for it is evidence. Next week, PSG play a match I circled in the calendar long ago. I will not write before watching. I will not conclude before counting. I will open the notebook, divide the pitch into twelve zones as always, and record every player position while the opponent has the ball. If the match repeats the pattern I saw across fifty-seven viewings, I will have more evidence. If it does not, I will have to rewrite my hypothesis from scratch. And if, after watching, all I have is one word — empty — then I will close the notebook and write exactly one line. Not enough. That line is not a conclusion. It is a promise to the reader that I will come back when I truly have something to say.