When the Tracking System Falls Silent

In the summer of 2026, at the Marseille sports magazine newsroom, I received a brief: write eight hundred words on the home team's pressing under Rudi Garcia. The editor-in-chief wanted it filed the same day. I shut my office door, spent the whole afternoon rewatching footage of twelve consecutive matches, and produced three thousand words circling a single fact: André Zambo Anguissa had made 127 ball recoveries in the opponent's defensive third. No newspaper in the city had counted that before me. The piece was shared more than fourteen hundred times — more than any official magazine article that year. I understood that data, placed correctly, can become the raw material of poetry. But I also remember another detail from that afternoon. When the paper came out, a senior editor called me in and asked bluntly: "What are you trying to prove?" I told him I just wanted to count. He shook his head. "Nobody buys a paper to read someone counting." Eight years on, looking back at the analytical apparatus football has built, I see a very different paradox. The problem is not missing data. It is too much data, enough to lull itself to sleep. How this industry operates deserves a pause. Every Ligue 1 club now has at least one analytics department with dozens of tracking cameras, every player wears a sensor chip, and every pass is logged to the centimetre. In March I visited the data centre of a mid-table club. On the big screen, hundreds of metrics ran like a symphony without a conductor. Yet when I asked the head coach which data he used to pick his lineup, he laughed and said: "I use my eyes and my gut." That answer does not reject data. It is a confession. A modern match can be described as a vast system of equations. Analytics hands us variables of astonishing precision: xG, PPDA, progressive passes, expected threat, packing rate. A top club can know exactly how many times its left-back lost the ball in the final thirty metres over the last ten matches. Yet the system still misses one unknown: the moment. The moment cannot be measured by a sensor. I have watched many Marseille matches at the Stade Vélodrome over twenty years. What I notice is this: the best matches I have ever seen are not the ones with the highest metrics. On the contrary, many matches that dominate the data end in disappointment. In the 2026-2026 season, Marseille had one game with 68% possession, 24 shots, and 2.7 xG — and lost 0-1. The press called it "football's injustice". I disagree. The visitors scored once, from a counter-attack in the 88th minute — but that was the only goal engraved in history. Meanwhile, the night Milan destroyed Barcelona 4-0 in 2026 — the match I consider the most perfect symphony of club football — needed no xG to prove its greatness. People remember Savićević, they remember Massaro. Nobody remembers Milan's possession figure that night. In the 2026 Champions League final in Istanbul, Liverpool trailed Milan 3-0 at half-time. Liverpool's possession then was 47%. But the moment Steven Gerrard headed in to make it 1-3 in the 54th minute changed the entire match — and no algorithm predicted it. Football is a game of moments, not of averages. So why does the analytics industry keep swelling? The answer lies in the market. The transfer market is a match with no referee, where every number is a free kick. When a club spends 80 million euros on a 22-year-old, it needs a reason to justify the outlay to its board, its shareholders, its press. Data becomes a legal shield. "He averages 8.4 metres of progressive carrying per touch" — a far safer answer than "we trust our eyes". But data does not score goals. It knows where the ball will go — only, sometimes, it knows wrong. I once witnessed a textbook case. In the 2026-2026 season, a young Brazilian striker was valued at 35 million euros on the back of outstanding xG and shot-creating actions in the Austrian Bundesliga. A French club signed him. Eighteen months later, he had scored three goals and was sold for eight million. The metrics were not wrong — they simply did not measure what needed measuring: the ability to adapt to Ligue 1's tempo, to withstand the pressure of the stands, and the loneliness of a twenty-year-old dropped into a foreign city. Over fifteen years of tracking transfers in France, I have counted at least forty cases of clubs paying high fees for players based solely on one season's data. The success rate among them does not exceed one third. There is a counter-intuitive reading of analytics' failures. People often blame data for being one-sided. But the problem lies elsewhere: football has never built a mechanism to audit itself. Metrics are created, used, then quietly die when they no longer fit the era. xG was born in 2026 to answer one specific question about chance quality. Twelve years later, it is used to judge everything — including things it was never designed to measure. That is the phenomenon I call "metric inflation": the tool of measurement becomes the goal. Medicine has drug regulators. Aviation has air safety agencies. Football has private companies selling data to clubs, and no one checks their methodology. If an xG algorithm is built on Premier League data, is it valid for the Swiss Super League? No one answers. No one is obliged to answer. The paradox lies elsewhere too. Advanced metrics are built on historical data — on what has happened. But football changes every season. Ligue 1's pressing in 2026 operates quite differently from the 2026 version. Average defensive line height has risen by three metres. Long passes have dropped 18%. Predictive models trained on old data gradually lose accuracy — yet no one has the courage to admit the tool they use is outdated. Better to keep a familiar metric than face the emptiness of not knowing. The most worrying thing is not that data is wrong. The most worrying thing is that we have forgotten how to interrogate it. Back to that summer afternoon in 2026 in Marseille. When I sat counting Anguissa's every recovery, I was not trying to prove that pressing matters more than individual technique. I was answering a very specific question: what is actually happening on the pitch that the human eye does not see? Data is a tool for answering the question, not an answer in itself. A football team, properly understood, is a system of equations that knows how to run — eleven unknowns that are never separate. But that system only means something when we know which problem we are solving. An empty stadium is a mirror: it does not reflect spectators, it reflects the loneliness of the game. And I believe every good analytical system must, in the end, aim to soften that loneliness — not amplify it with numbers no one reads.