11.8 Kilometres a Match and Four Months Off the List: The Limits of a Model

On 12 January 2026, at 9:14 in the morning Copenhagen time, I opened my inbox and read a short note from the club's sporting director: the Senegalese defensive midfielder I had recommended for signing in July 2026 had been removed from the squad list for the second half of the season. On my second monitor the spreadsheet was still open. One cell read 11.8 km, the average distance he covered per match. Another read 6.2, his number of ball recoveries. The third read 4.1, the midfield duels he won per 90 minutes. I stared at those three cells for a long time while snow settled on the roof of the building across the street. Not one of those cells was wrong. He really did run 11.8 kilometres a match. He really did recover the ball 6.2 times. What I misread was a row the spreadsheet never had: he needed a dressing room that spoke his language, a fixed role in a system that would not keep changing, and a coach who would believe in him long enough to wait. Four months after signing he had 214 minutes in the Superliga, two substitute appearances, and not a single start in the final three months. I tell this story in the voice of a man who has just lost a bet he rigged himself. My trade is reading football through numbers. For most of the year I cover badminton for the Danish market, where a rally lasts forty seconds and every touch of the racket leaves a clear trace. But whenever the transfer window opens, I pick up a few consulting contracts with Nordic football clubs. That is the part of the job that costs me the most sleep, because the transfer market is the only place where a beautiful dataset can be sold as a promise. This winter in Denmark, clubs walked into the window with the same two questions. They needed someone who could play immediately, or someone they could sell for three times the fee. Very few buyers say the third thing out loud, the thing everyone knows: they need a person who will not break the dressing room. Our model had no field for that. The club I advise belongs to the group that treats data as a mother tongue. In Denmark that model has a famous example: FC Nordsjælland, tied to the Right to Dream academy and champions of the Superliga in 2026-23, the second title in the club's history after the 2026-12 triumph. Names such as Mohammed Kudus, Kamaldeen Sulemana, Ernest Nuamah and Simon Adingra all came through that pipeline before moving to bigger leagues. The story is usually told neatly: the algorithm finds the talent, the talent shines, the club cashes in. What gets left out is the first months, when a twenty-year-old from Accra lands in a coastal Nordic town in January and learns to live through darkness at four in the afternoon. In 2026, as a student, I calculated PPDA for FC Nordsjælland across thirty matches. They pressed harder than the rest of the league, allowing 8.5 passes per defensive action, 2.1 below the league average. They finished seventh. The panel grading my thesis said it read like stale bread. I sat alone in a cafe afterwards and wondered why a fact that clear moved nobody. PPDA does not measure the heart, but it points to where the heart is beating. A second lesson arrived this October, and it cost far more. His file was built over eighteen months, from forty-two matches in the Senegalese top flight, a handful of continental cup games, and video sent by local stringers. I normalised every metric into percentiles and adjusted for the strength of each league. Across those forty-two matches he sat in the 96th percentile for distance covered, the 91st for ball recoveries, the 88th for midfield duels won. An average of 11.8 kilometres per match put him in a group only four midfielders in the Superliga reached the previous season. I presented the file in a meeting that lasted fifty minutes. Seven people were in the room. A veteran scout, twenty years of watching football in Africa and Eastern Europe, raised his hand after I switched off the projector. I still have his sentence written down word for word: "He will need six months to learn how to live here before he learns how to play. If we do not have those six months, do not sign him." I nodded, and in the recommendation section I wrote that the integration risk was low. He trained well. His first fitness tests beat expectations. A four-year contract, a salary around sixty per cent of the first-team average, and a release clause the board believed would let them sell for triple the fee after two seasons. On paper it was a tidy deal. In August he came on for twelve minutes on the opening day and made three recoveries. In September, after four winless games, the head coach switched from a 4-3-3 to a 3-5-2. The lone holding role disappeared, replaced by two shuttlers who had to handle the ball in tight spaces under pressure. That was a player my file had never described, because my file had described him accurately. In October he strained a hamstring and missed three weeks. In November two senior midfielders returned from injury. In December he trained only. In January his name came off the list. Based on my experience tracking matches in the Superliga this season, I noticed something the spreadsheet will never rank: a person's speed of adaptation is not measured in kilometres but in the number of evenings spent alone in a rented flat, and in the number of times he picked up his phone without knowing who to call. There is a trap people in my trade fall into. We believe that if a metric correlated strongly with success in the past, it will manufacture success in the future. High distance covered correlates with the quality of a defensive midfielder, until it correlates with a team that keeps losing the ball and a player who has to chase it. The same data, two readings, two opposite conclusions. In 2026 I made a similar mistake. During Denmark's group-stage match against France at the World Cup in Russia, I wrote that the national team pressed chaotically because their PPDA was very low. A former international read the piece on television and asked me straight: "Have you watched the tape?" I rewound it fourteen times in the editing room until three in the morning and realised I had ignored the defensive positions and the purpose behind the whole team's pressing. I rewrote the piece in two versions, one by numbers, one by eye. Numbers only recount the past, while football lives in the future. That is why a good model must always leave a gap for what it cannot measure, and that gap is usually where people live. The transfer market runs on a paradox. Models rate youthful potential extremely highly, because potential can be expressed as a percentile and sold for money. Models rate dressing-room chemistry extremely low, because chemistry can only be heard with the ear, and nobody pays for an index they cannot resell. So the cleanest signings on the spreadsheet are often the most fragile in reality. One thing I want to state plainly, because people in this trade keep confusing it. I do not believe in luck; I believe in what luck conceals. He was not cut because of a curse, but because of four concrete decisions: a system that changed, an injury that arrived at the wrong moment, a limited foreign-player slot, and a model with no room for a variable called loneliness. Three of those four were predictable. We simply chose not to predict them. In the report I sent the board this week, I added a new column to the model and named it Column K. Column K holds no percentiles, no kilometres, no goals. It holds three questions: does this player have anyone here who speaks his mother tongue, is the club willing to wait six months, and will the person who recommended him carry the responsibility if those six months stretch to twelve. I know a model with Column K will produce fewer names, and some of them will be less glamorous. Next summer a different file will land on the table, with numbers even prettier than this one. I have told myself I will read it more slowly, spend a few more phone calls, ask a few questions that no algorithm can answer. Viewers see the goal; I see the sequence of events before it. But that sequence began on an evening in Dakar, when a young man called home and nobody at the club knew. Will I have the nerve next time to write his name into Column K, or will I leave it blank again?