The Heat Map Cannot See the Match

On a Tuesday morning at the training centre, I stood on the second-floor balcony next to a coffee machine that had been broken since last season. Below, the team walked out for their warm-up. On the analysis-room screen, a heat map had just synced. It showed their central midfielder covering almost the entire left channel. The deepest red zone sat right in front of the opponent's penalty area. On paper, he was the man who had created the most danger in the match. I had sat in the stands for that game. That player barely set foot in any meaningful red zone. He moved constantly, true, but mostly in short steps to keep distance from the full-back, receiving the ball in safe positions and returning it. The heat map records presence. It does not record intent. That moment made me think of something that has haunted me for many seasons. We live in an age where every action can be reduced to a number. Every player can be measured by a metric. Every match can be summarised in a three-page data sheet. But the real match, with its breathing, its glances, its decisions made in half a second, still sits somewhere outside the map. When the press-room door closes, I begin to hear the match more clearly. I remember 2026, when I was twenty-seven and assigned to follow a major French club. My debut match was a home defeat. Afterwards, a veteran reporter stopped me in the corridor: "The dressing room is no place for a girl." That week, instead of arguing, I stood in the hallway taking notes for four hours. I watched a midfielder leave the pitch in the sixty-second minute without looking at anyone. I read his frustration only from the way he kicked a water bottle. No metric recorded that. No analytical sheet detected it. Every season is a heartbeat, and I am only trying to catch the right beat. Now, more than twenty years after my first day at a local radio station, I still keep that habit: standing at the edge, watching the unknown, and trusting what I see over what is printed. This piece is a record of that paradox, of the gap between data and the match. Fifteen years ago, the analysis department of a mid-table French club had two people and one old computer. Today, that same club has a full unit of dozens of specialists, outsources data match by match, and buys motion-tracking packages at twenty-five frames per second. The revolution arrived so fast that many head coaches still have not learned to read what it delivers. People began calling xG a measure of justice. They began comparing two teams by possession. They began evaluating a striker by expected goals per shot. Data companies sprang up like mushrooms, each claiming the most accurate model. Big clubs hired physicists and mathematicians. Leagues began using semi-automated technology to determine offside. Everything became measurable, and people believed that what can be measured is therefore true. I do not oppose data. I oppose using data to replace the eye. There is a difference between a tool that helps us ask better questions and a tool used to conclude on our behalf. Many clubs crossed that line without noticing. In the press room, they ask the coach about the possession figure. He answers that his team controlled the game. But controlling the ball and controlling the match are two different things. A team can hold seventy percent of the ball and still be led by the nose. A team can hold thirty-five percent and still own every important moment. The number cannot tell the difference, because it only counts the time the ball spent at whose feet, not the time someone was making a decision. The postman never asks me what I need; he simply leaves an envelope in silence. The postman in my story is a staff member at the training ground who told me things no report contains. That players hide sweets in their jacket pockets when leaving the hotel. That an older player is always the last to leave the dining room. That a young newcomer did not dare sit at the same table as the seniors for three weeks. Those details, added up, say more about a team's state than every fitness sheet I have ever read. The Moscow night taught me that the truest sources rarely carry a business card. I tell these stories not out of nostalgia. I tell them to place two ways of seeing side by side, and to show that the humbler second way is not obsolete. The problem is that the first way is increasingly occupying public space. Start with the heat map, the most used and most misunderstood tool. In essence, it is a way to visualise positional data. Each frame, the system records the coordinates of every player on the pitch. Add all the points, smooth them, colour them by density, and you have a picture of where a player has been. That is all. It does not say what he did, with whom, under what pressure, or under whose instruction. A central midfielder who stays near the centre circle may be doing exactly his job: holding position, screening, organising. His heat map will be a small blob, looking rather poor. Another midfielder running everywhere like a shuttle will have a brilliant heat map, even though most of that time he was chasing the ball or compensating for his own misplaced positioning. If you look only at the heat map, you will conclude the second player is more dynamic, more influential. That conclusion can be entirely wrong. I once saw a more extreme case. A full-back whose heat map stretched across the whole channel, from his own goal line to the opponent's touchline. Looking at it, one would think he was a marauding weapon. In reality, in that match he was repeatedly pulled out of position, exposing the space behind, and the team conceded two goals from that channel. The heat map does not colour mistakes. It only colours presence. This is why I say the heat map has become a new kind of divination. Like the ancients reading tea leaves to guess fate, people today read red clouds to guess a player's value. Both rely on an unverified belief: that a trace left behind fully reflects the essence of what happened. In both cases, the interpreter matters more than the thing interpreted, and the interpreter often does not realise he is imposing meaning on an image that is inherently neutral. I am not saying the heat map is useless. It is useful for spotting broad trends the human eye struggles to synthesise across ninety minutes. It is useful for noticing that a team is heavily tilted to one flank, or that a line is being repeatedly carved open. But it must be read as a question, not an answer. Whoever reads it must ask: why was he there, and what happened in the moments the map did not record. Next is expected goals, abbreviated xG. It is a model estimating the probability that a shot becomes a goal, based on position, angle, shot type, and some contextual factors. It changed how people evaluate a match. Instead of only looking at the scoreline, people look at which team created better chances. That has real value, especially with a large enough sample. But xG is abused no less than the heat map. The problem is that the model measures the quality of the final shot, not the quality of the process leading to it. A team can create a gilt-edged chance through a thirty-pass move of pure beauty, and can create an equivalent chance through an absurd defensive error. With xG, the two chances are worth the same. Moreover, xG cannot measure decision speed. A striker can stand in a high-xG position, but only because he took too long to control the ball and the defender recovered in time. Another striker may have lower xG because his decision was so fast that nobody could react, and the goal came from a situation people thought was nothing. The model does not see that speed. It only sees position and space calculated before the ball left his foot. On top of that, xG depends on the model's training data. Each provider has a different model, producing different results for the same shot. This is rarely stated on commentary shows. Viewers are given a single number as though it were objective truth, when in fact it is the product of a series of model choices the viewer is not told about. To me, a number from a model that is not clearly presented has significantly reduced value. Then comes the pressing metric, usually abbreviated PPDA, the number of passes the opponent is allowed per defensive action. The lower the figure, the more fiercely the team presses. This is a genuine contribution of data, turning something very hard to count by eye into a simple number. But like any metric, it only measures surface behaviour. It does not distinguish pressing to win the ball from pressing to push the opponent into a prepared zone. It does not distinguish organised pressing from the whole team charging forward in panic. I once followed a team across three consecutive matches. Their PPDA dropped sharply, from average to high. The press wrote that the coach had changed his style, that the team was playing with more fire. The reality was far simpler: they conceded early in all three matches, and high pressing was the reaction of a team chasing the game, not a tactical choice. The number was right. The story was wrong. Someone who reads only the number will write the wrong thing about the match, and the worrying part is that this way of writing is increasingly common. From here a larger problem emerges: dependence on the average. Data always tends to pull everything toward a representative value. A player who runs an average of eleven kilometres per match is considered hard-working. But if ten of those kilometres were walking to take up positions, and one kilometre was three decisive sprints, then eleven added together says nothing about value. Conversely, a player who runs nine kilometres, including seven sprints at key moments, may be the most important man on the pitch. The average erases the distribution. It turns a volatile match into a single point. And when people compare two players by a single point, they are comparing two photographs, not two films. For a sport where everything happens along a stream of time, using photographs to conclude is a methodological error. I once lived through a season in which data and results clashed bitterly. A team played well, created many chances, but kept dropping points. The analysis department concluded they were unlucky, that they only needed patience. By January, the coach was sacked. People said the data had warned that everything would be fine. It did not warn that two groups of players in the dressing room had stopped looking at each other. It did not warn that a few individuals had lost faith in the man in charge. Data saw the shot. It did not see the person. Across a decade of following clubs, I learned that a collective's psychological state is the most important and hardest-to-measure variable. It appears in no model. It appears in small details: who claps before stepping onto the pitch, who stays silent at meals, who is the first to stand when training ends. That is my data, and it sits in no database. What is striking is that the most successful clubs I have observed are not the ones with the most data. They are the ones that know how to use data to ask questions, then answer them with the human eye. They use numbers to know which clip to rewatch. They do not use numbers to conclude before watching. One example stays with me. A famous coach once instructed his analysis department not to deliver any conclusion until every assistant had watched the match twice, once following the ball, once following a single chosen player. He said their data was very good, but it should only appear after the human eye had finished its work. That approach made the number a servant, not a master. By contrast, I know another club that made transfer decisions almost entirely on computer models. They signed a striker with outstanding expected-goal output in a minor league. He failed completely. No one at the club had ever gone to watch him live, ever spoken to old teammates, ever checked how he lived, what he ate, how he slept. The model was right in numbers. It was wrong in humanity. This is the point I want those in my profession to remember. Data is a language. And like any language, it has grammar, vocabulary, and things it cannot express. However skilled the speaker, he cannot use it to tell a story for which it has no words. A team's match always contains things outside the vocabulary of any model. I came to this profession from local radio, where there was no analysis department, no data package, only a microphone and a match to narrate. There I learned to trust my ears. I learned that the roar of a stand, the shout of a coach, the sigh of a player leaving the pitch, are all data. They are rough, hard to systematise, hard to sell, but they are honest in a way a spreadsheet cannot be. Now, working with clubs that have hefty analysis units, I keep that habit. I spend more time in corridors than in press rooms. I talk to bus drivers, to waiters, to security staff. I write nothing down in those conversations. I just listen, and let them settle. When it is time to write, I realise those small stories are what help me tell a true story from a story data wants to tell. There is a paradox that always troubles me. Data helps a club understand opponents better, but it does not help a team understand itself. A coach can know the opponent's model inside out, know which flank they prefer, but he cannot know that his key player is struggling at home, that he is not sleeping, that he is thinking of retiring early. Those things are in no report. They surface only when someone is close enough to notice. That is my job, and I believe in it. Being thrown out is how the organisers gift me a different angle. If I were allowed into the dressing room, I would witness what everyone witnesses. Because I was pushed outside, I had to develop another way of observing: looking at footsteps, at how a cup is held, at the glance of someone looking down. That way does not replace data, just as data does not replace it. But it gives me something data lacks: context. Context is the thing most missing from every online football argument today. People argue with numbers torn from context. They compare two players without knowing that one just returned from injury, while the other is playing out of position. They conclude a coach is incompetent without knowing he just lost three pillars to suspension and is forced to use reserve-team players. A number does not speak for itself. People make it speak, and make it speak the way they want. This is where I want to speak plainly about the back-three trend, celebrated as a leap in modern football thinking. From following many clubs, I see that most switches to a back three are not the product of a new vision. They are defensive reactions. A back four is carved open for a few matches, the coach needs a risk-reduction solution for his own position, and the back three appears as a shield. It is not always so. Some teams use a back three deliberately, to free two wing-backs high, to squeeze the opponent into the middle. But in many other cases, it is a way of telling the board that something was done under public pressure. The back three in those cases does not solve the tactical problem. It only hides it a little longer, so that if it fails, the coach has a reason to explain that he tried to change. I know this view will upset many. Modern football has its own vocabulary to praise flexibility and tactical thinking. But real flexibility and flexibility for self-protection differ in the final result, and the final result is rarely discussed in glossy analyses. Similarly, the story of upsets in cup competitions is always told as a fairy tale. A small team beats a big one, and the media calls it the magic of football. From following such matches, I see they are usually the result of two very concrete things: the strong team rotating because of a crowded calendar, and the weak team executing high pressing with discipline. There is nothing magical in that. There are priority choices, there are fitness calculations, and there is a weak team willing to suffer for one single result. When I write this, people often ask whether I am stripping football of its poetry. I think it is the opposite. Understanding an upset correctly is more exciting than calling it a miracle. A miracle cannot be repeated by anyone. A process can be learned. Telling the story of planned effort, of a weak team that knows exactly what it must do, of a coach willing to bet on lesser-known players, is the way to respect both the winner and the loser. And this is what I believe is the core of everything. Every tool, whether a number or an eye, is valuable only when it helps us understand people better. If a heat map makes us see a player as a data point, it has failed. If a model makes us forget that behind every shot is a person trying to redeem a mistake, it has failed. The success of data lies not in how many judgements it replaces, but in how many right questions it helps us ask. As the seasons pass, I realise the pieces that live longest in readers' memory are not the ones with the most tables. They are the ones about a player quietly straightening the substitutes' bench before leaving the pitch, about a coach standing alone on the touchline until the stands are empty, about a security guard who still remembers the name of a player who left the club ten years ago. Those details cannot be modelled. Because once you model them, you have already lost them. What I want readers to take from this piece is not a rejection of data. I do not want anyone to abandon numbers. I want everyone to read a number with a question: what does it count, what does it overlook, who decided the model, and what is happening that it cannot see. Once you can ask those questions, you will no longer be led by the number. You will use it as a craftsman uses a tool, not as a believer praying before a charm. In the coming weeks, as the season enters its decisive phase, I will watch a few specific signals. First, I follow how coaches rotate when the calendar is crowded. Whoever dares to rest key players is playing a long-term gamble, and I want to see whether they have the patience to pay for a few temporary defeats. Second, I notice teams that switch from a back four to a back three after a bad run. If the change comes after two defeats, it is almost always a defensive reaction, and I will write about it as what it is. Third, I listen to what substitutes say at press conferences, because those asked least often tend to speak most truthfully. I will also keep doing what I have done for years: standing at the edge, drinking coffee from a broken machine, and trusting what I see. The stadium is empty, yet I still hear the applause of thousands from memory. And in that applause, I hear what no data sheet records: that football, in the end, remains a story of people, told by people, for people. The thought I want to leave here is a question for myself, and for those who do my job. As our tools grow stronger, will we still have the courage to say they are not seeing the most important thing? And as readers grow used to tidy numbers, will we still have the patience to tell messy, contradictory stories that cannot be summarised in a spreadsheet? If the answer is yes, this genre still has a place. If the answer is no, we have not lost a match. We have lost the ability to see it at all.