For more than a century, football was told as a story of talent, instinct and a good eye. You loved a player because you felt it. You signed a striker because an old scout, with a hundred matches in his legs, nodded at the touchline and said: that one, he has something. The decision was made in the gut, never in a spreadsheet.

A fact everyone looked at without seeing

That belief had something reassuring about it. Football was an art, and art cannot be measured. Putting numbers on the beautiful game seemed almost a lapse of taste, a betrayal of its magic. The most respected leaders in the sport were convinced of it, and that conviction structured everything: how to recruit, how to pay players, how to judge a manager, how to tell the story of a win.

And yet a very simple fact lay in plain sight, without anyone thinking to stop at it. A football match is a sequence of actions, passes, shots, runs, balls lost and won back. Thousands of events, each observable, each recordable. The question that would change everything had simply never been asked seriously: what if all of this followed regularities? What if the apparent randomness of the pitch hid laws one could learn to read?

That question took seventy years to find its true answers. It is, today, transforming a sport once thought impervious to calculation.

A notebook and a pencil at Swindon, in 1950

The story begins with a tiny gesture. On 18 March 1950, on the pitch at Swindon Town, a spectator loses patience. The match against Bristol Rovers is laboured, the attacks repeat and lead to nothing. This man is called Charles Reep. A Wing Commander in the Royal Air Force, an accountant by training, methodical to the point of obsession. At half-time, he takes a notebook and a pencil from his pocket and starts to note. Every possession, every pass, every shot. Where the ball goes, where it is lost, what precedes a goal.

No one around him grasps what is happening. Nor does Reep, probably. He just wants to understand why his team attacks so much and scores so little. But that day, at Swindon, is born the first serious attempt to turn a football match into data.

The state of the art at the time was memory and intuition. A manager relied on his recollection, inevitably partial, inevitably biased by the last spectacular moments. Reep introduces a radical idea: what if we replaced memory with recording? Over forty-five years, he would note by hand more than two thousand two hundred matches. Hundreds of thousands of passages of play, set down on squared paper. A monk's labour, carried out without computer, without video, without the slightest modern tool. Just the eye, the hand, and an accountant's patience.

Reep's conviction is that football has patterns and that these patterns can be identified. Reep demonstrates that the game, beneath its appearance of chaos, contains regularity. He calculates, he compares, he looks for the schemes that precede goals. He is the first to treat the pitch as an object of study rather than a spectacle.

The price of a good idea misunderstood

Reep could have become a hero. He long remained a controversial figure, and that part of the story deserves to be told honestly. For from his thousands of records, he drew a conclusion: most goals come from short sequences, of fewer than four passes. He deduced that the ball should be sent forward as quickly as possible, minimising patient build-up.

A few English managers listened to him and built on his numbers a direct, physical football, without detours. The results were sometimes spectacular. But the interpretation was partly wrong. Reep had seen correctly on an immense point, the existence of measurable regularities, and had been mistaken about what they meant. The football world, for its part, remembered only the mistake. For decades, statistics were associated with the long, ugly game, and the very idea of counting remained suspect, almost disqualified.

It is a mechanism the history of science knows well. A sound intuition, carried too early, poorly equipped, produces a shaky conclusion, and that conclusion discredits the intuition as a whole. We would have to wait for the tools to catch up with the idea for it to be finally understood.

The question no one had thought to ask: not all shots are equal

The conceptual click comes in April 2012, in a blog post. An analyst named Sam Green, who works for the statistics company Opta, publishes a seemingly innocuous piece on Premier League scorers. He asks a question of insolent simplicity: why do some strikers need twice as many shots as others to score the same number of goals?

The answer holds in one phrase that others, in North American hockey, had already formulated in their own way: not all shots are equal. A header at the penalty spot after a corner has nothing to do with a clear chance launched on the counter-attack. What counts is not the number of shots, but their quality.

Green turns this obvious truth of the pitch into a number. From thousands of strikes, he assigns each shot a probability of ending up in the net. He takes into account distance, angle, the type of action, the part of the body. This value he calls the expected goals, abbreviated xG. A difficult shot is worth 0.05 expected goals. A one-on-one is worth 0.4. Added up over a match, these figures say how much a team should have scored given the quality of its chances.

This innovation is remarkable because it finally reconciles the number and the sense of the game. Where Reep counted sequences, Green measures chances. He gives decision-makers an instrument that captures something the score alone hides: a team can lose while having played better, win while having been worse. Football is a low-scoring sport, and therefore soaked in chance. xG lets you see beneath the chance, to distinguish real performance from mere luck. That is exactly what a decision-maker needs: a reliable signal where noise dominates.

A physicist at Liverpool

An idea, however beautiful, is worth only what you make of it. That same year, 2012, in Liverpool, a discreet man enters the club's history through a side door. His name is Ian Graham, he is Welsh, he holds a doctorate in theoretical physics from Cambridge and he is nothing like a football man. The club's new owners, an American fund that also owns a baseball team, the Boston Red Sox, believe that what worked on baseball diamonds can work on English pitches. They give Graham a mission simple to state, immense to achieve: to build the first true research department of a major club, and to base decisions on models rather than impressions.

The beginnings are rough. Managers are wary, the old culture resists, a single player who refuses to buy in is enough to sink a season. Then the pieces fit together. Graham and his team are not content with xG. They build their own measure, the possession value, which assigns a value to each action, a pass, a tackle, a movement, asking at every instant: by how much did this action change the probability of scoring?

The moment that will endure is the summer of 2017. The manager, Jürgen Klopp, wants to sign a talented German winger, Julian Brandt. Graham's models point to another name, a player coming off a quiet season in Italy whom few consider a star: Mohamed Salah. The numbers say that Salah generates, from the flanks, far more dangerous chances than an ordinary winger. The club trusts the analysis. Salah will become one of the best players on the planet.

The rest is well known. Virgil van Dijk, Sadio Mané, Alisson, a spine largely lit by data. A Champions League in 2019, an English league title in 2020, the first in thirty years. What makes this success remarkable is not that data replaced flair. It is that it disciplined it. Klopp watched videos and felt the players, like any great manager. But every name on his list had first passed the filter of the models. Intuition and calculation, at last, worked together.

A third-division club that refuses to play the rich man's game

While Liverpool deployed giant means, a parallel story, more improbable still, was unfolding in west London. Brentford, a small third-division club, was on the edge of financial ruin. In 2012, a childhood supporter takes control of it. Matthew Benham, also a physicist by training, a former trader turned professional gambler thanks to statistical models that beat the bookmakers more often than not.

Benham applies to the club the principles that made his fortune: seek out inefficiencies, buy what the market undervalues, sell what it overvalues. Where the big clubs pay dearly for known names, Brentford unearths players forgotten in the lower divisions, grows them, then resells them at the peak of their value. The figures are dizzying. Ollie Watkins arrives for less than two million pounds and leaves for nearly thirty. Neal Maupay comes in at one point six and goes out at twenty. Saïd Benrahma, one point five to buy, twenty-five to resell.

The club even closes its academy in 2016, judged inefficient, to create a B team made up of youngsters rejected elsewhere. In 2021, Brentford is promoted to the Premier League for the first time in seventy-four years, with one of the smallest budgets in the elite. The feat is remarkable because it owes almost nothing to money and everything to a more accurate reading of the market. Benham understood before the others that football brims with emotion, and therefore with irrationality, and therefore with opportunities for whoever keeps a cool head.

What really changed

What played out between Reep's notebook in 1950 and the titles of the 2020s goes far beyond sport. It is a change in the status of football itself. For a century, it had been treated as a pure product of talent and chance, a domain where only the sensitivity of the trained eye mattered. Data tips it into another category: that of systems one can model, measure, and therefore improve methodically.

The right words may be those of one leader in the field: where there is a lot of emotion, there is necessarily a lot of inefficiency, and therefore a lot of opportunity. Football has not ceased to be an art. It has simply become, in addition, a market that some have learned to read better than others.

This shift recalls another story, born in another sport. In the early 2000s, a cash-strapped baseball team, Billy Beane's Oakland Athletics, held its own against the richest by recruiting on statistics the rest of the league disdained. The book that told this adventure, Moneyball, became the manifesto of a whole generation of leaders. It is no accident that Liverpool's owners come, precisely, from baseball. The breakthrough did not spread from one stadium to another by chance. It travelled from one sport to another, carried by people convinced that a single idea applied everywhere: where everyone decides on feeling, whoever measures gains the edge.

A slow innovation, then sudden

The story is a textbook case of the mechanics of real innovation. Not the solitary genius who changes the world overnight. The slow sedimentation, punctuated by breakthroughs that only become visible in hindsight.

Between the Swindon notebook and the first title lit by data, more than sixty years passed. Reep had the intuition but not the tools. Green had the right idea just as a company like Opta was finally mapping every event of a match. Graham and Benham had, on top of that, the computing power and the institutional will to bet everything on it. Each generation inherited the underground work of the previous one, long without knowing it.

The general public saw nothing coming. Expected goals stayed confined to analysts' computers until 2017, when a flagship British television programme began to display them on screen. In a few seasons, a concept born in an obscure blog post found itself in the mouths of commentators and supporters. What seemed esoteric had become obvious.

What the numbers will never tell

It would be easy, and false, to conclude that the machine won against the human. The best practitioners in the field are the first to recall the limits of their art. A model measures probabilities, not courage. It captures neither the fear of a young player on a big night, nor the alchemy of a dressing room, nor that extra something that makes a team transcend itself. xG themselves see only part of the game, the shots, and ignore a thousand gestures that win matches.

That is perhaps the most beautiful part of this story. Data did not kill the magic of football. It made people humbler before what they did not understand, and more lucid about what they could finally understand. The best clubs are not those that replaced the scout with the computer. They are those that made the two speak to each other, the eye and the model, the gut and the spreadsheet.

What this story says to any company that innovates

At bottom, this account does not belong to football alone. It describes what happens to any field where decisions have always been made on instinct, convinced that calculation has no place there. Research, recruitment, investment, product design: wherever emotion and habit dominate, there hide inefficiencies no one sees because no one measures them. And in those inefficiencies sleeps a considerable advantage for whoever dares to look differently.

Innovation, in all its aspects, always follows this mechanism. It begins with a fact no one notices, continues with a pioneer mocked in his time, crosses decades of patient work, then suddenly imposes itself as a retrospective obviousness. Those who understand the breakthrough while it is still invisible build an advantage that others will take ten years to close. Reep noted alone in the stands when everyone found him eccentric. Today, no major club recruits without a model. Between the two, there was no miracle. There were a few stubborn people, a notebook, a pencil, and the courage to count what everyone believed uncountable.

To go further

  • To read: How to Win the Premier League by Ian Graham (2024), the inside account of the data revolution at Liverpool, by the man who led it.
  • To read: Expected Goals by Rory Smith and Net Gains by Ryan O'Hanlon, two complementary histories of how numbers conquered football. And of course Moneyball by Michael Lewis, the ancestor from baseball.
  • To watch: the talks by Ian Graham and Matthew Benham filmed at the MIT Sloan Sports Analytics Conference.
  • To follow: the new frontiers of analysis, the tracking data that records every movement of every player twenty-five times per second.