The Data That Flatters

A soccer club spends a decade building the most expensive analytics department in its history and ends up in a relegation fight with a roster full of sprinters who cannot pass. This is not a scouting failure. It is what happens when numbers get hired to agree with a decision someone already made upstairs.

A North London club’s technical director reportedly favours a metric that combines speed, endurance, and explosiveness. Most of their recent signings score ninety or higher on it. Only two of their regular starters rank inside the English top flight’s top hundred and fifty passers, in a sport where roughly four hundred and fifty passes change hands every match. The squad runs beautifully. It just does not play.

The older version of this story sits in baseball, a couple of decades back. A small-market club, outspent by richer neighbours by a factor of three, decided to stop asking scouts whether a prospect looked the part and started asking whether he actually reached base. The orthodoxy at the time rewarded broad shoulders, clean swings, and a certain jawline. The new approach rewarded whatever correlated with runs, which turned out to be unglamorous things like patience and walks. The scouts were furious. The club won twenty games in a row. A book was written, then a film was made, and the lesson curdled into a slogan that every sport has since claimed to follow without actually following.

A West London club, bought a few seasons ago by an American private-equity consortium more familiar with baseball and basketball than with relegation, is running a different experiment with the same underlying faith. Ten managers in ten seasons, each hired to fix what the last one broke, each replaced before the system could discover whether he was right. The new owners describe it as a portfolio approach. From the outside it reads more like a man changing thermostats while his house fills with smoke.

What unites the two is not incompetence but confidence. Somewhere in each building, a document exists that explains why the current approach is working. The numbers are real, the slides are handsome, and the conclusions were written before the research began. The hard part of using data well has never been collecting it. It is being willing to read what it says when it says something you did not want to hear.

Most organisations, asked whether they would prefer a report that confirms their plan or one that complicates it, would claim to want the latter.

Most of them do not.