The Numbers Game

Everyone has an opinion about soccer. At the bar, in the family group chat, on the car radio, everyone knows why the team lost, who should have started, and which coach needs to go. What Chris Anderson and David Sally propose in The Numbers Game is to stop and ask: which of these opinions holds up when we look at the data?

The authors are two university professors, one in operations and the other in game theory, who are also fans of the sport. Inspired by the success of baseball analyzed through numbers, the story popularized by the book and movie Moneyball, they wanted to know whether soccer also hides counterintuitive truths.

The answer is yes, but with care. The book shows that many beliefs repeated for decades are wrong, and also that soccer is hard to measure. Not everything that gets counted matters, and not everything that matters gets counted. The fun lies in learning to tell one from the other.

One of the most curious characters in the book is Charles Reep, a Royal Air Force officer and trained accountant, who decided to analyze soccer matches systematically, writing down every play by hand. He recorded thousands of matches starting in the 1950s, with an almost obsessive dedication.

His conclusions seemed strong: the vast majority of goals came from plays with few passes, usually three or fewer. From this he drew a recipe: to score more, it was better to launch the ball forward quickly, avoid long passing sequences, and try to reach the goal by the shortest route.

The idea influenced English coaches and directors and helped consolidate a direct style of play, the so-called long ball. It was the first great example of data analysis in soccer, and also an excellent case of how numbers can lead to wrong conclusions when misinterpreted.

Reep's mistake is subtle and worth understanding. He observed that most goals came from short sequences, but he forgot to ask how many short sequences exist in a game. The answer is that the vast majority of possessions end quickly, through a turnover, a shot, or a stoppage.

If almost all plays are short, it is natural that most goals come from them, without that meaning that short plays are more efficient. The right thing would be to compare the probability of a goal in each type of sequence, not just count where the goals came from. The authors show that, looked at from this angle, the advantage of direct play disappears.

It is a good lesson outside of soccer. When someone says that most satisfied customers came from a certain sales channel, it is worth asking how many customers came from it in total. A statistic without the right denominator becomes propaganda.