Playing by numbers

By Ed Smith

Ten years ago, walking past the Harvard bookstore, I saw a billboard inviting me to discover "What happens when sport is exposed to the scientific method". It was an advert for a book that was going to become famous: Michael Lewis's "Moneyball".

I was spending a semester at Harvard—right in the middle of my cricket career, as it turned out, though the number of grains of sand left in the egg-timer is never clear at the time. I stopped in my tracks in the November snow. That science could revolutionise sport: the idea was thrilling, and also troubling.

The intellectual in me, long frustrated by sport's suspicion of ideas, sensed an overdue balancing of the ledger, a renaissance of the nerds at the expense of the jocks. But the sportsman in me was heading the other way. I had been too rational in my early 20s, to the detriment of my game, and was now searching for my authentic voice rather than a scientific solution. I was trying to think less.

Today, despite the freedom and distance I now enjoy, despite all the fieldwork produced by professional teams pursuing data-led strategies, I am still torn about this. I credit reason when I have evidence; I appeal to intuition when I don't. This feels like good judgment. Intellectually, it is a mess.

Football, long to be found at the unreconstructed end of the spectrum, is now embracing statistics. We know there were 19 times last season when a player made 100 successful passes in a Premiership match, usually Mikel Arteta of Arsenal (6) or Michael Carrick of Man United (5). We know who made the most interceptions (Morgan Schneiderlin of Southampton, 139). Football was bound to succumb to stats eventually. It is lucrative, so any competitive edge translates into hard cash. But it is also, especially in Britain, resolutely blokeish. Three of England's past six regular captains in Test cricket have been graduates; it is hard to imagine football having even one any time soon. Few sports have a more ingrained contempt for rigour.

There is a further problem with using numbers to decode football. Baseball is a statistician's fantasy: every event is discrete and self-contained, producing a clear answer to a definite question. Ball or strike? On base or out? Fielding error or not? Football, by contrast, is a fluid sport with stretches of unbroken play in which the players interact subtly, and often imperceptibly. An arching run by one player, who never touches the ball, may provide vital space for a team-mate who scores the goal. We have to measure fleeting shadows as well as imprints in the sand.

Defence, too, is about what does not happen as well as what does. Paolo Maldini, the supremely assured left-back for Milan and Italy, made on average one tackle every two games. He seldom deigned to get his knees dirty because his positioning was so good. This reminds me of an argument I once had with a diplomat who was conducting a round of data-driven analysis of "British foreign policy". Isn’t the art of diplomacy, I asked, defusing situations so effectively that they leave almost no trace?

Football is also highly susceptible to luck. Its scoring unit, the goal, is so rare and clunky that randomness is built into the game: a cup final can be settled by a deflection. That’s why bookmakers pick favourites less often in football than other sports. Distinguishing between signal and noise is unusually difficult.

But the numbers people have not flinched from the challenge. In "The Numbers Game", published in May, two academics, David Sally and Chris Anderson, develop a series of powerful arguments about how football matches are really won. (Romantics, look away now.) First, defence matters more than attack. Goals that don't happen are more precious than those that do. Secondly, a team’s brightest star is less significant than its weakest link. Because footballers interact with each other so many times in a match, strategic togetherness trumps stand-alone talent.

Evidence-led strategy can trump conventional wisdom. At Euro 2012, Spain were widely berated for lacking adventure when they played an extra midfielder and no striker. When they beat Italy 4-0 in the final (pictured), with Cesc Fàbregas playing a full hand as the non-striker, their tactical conviction was as cheering as their skill. The tournament was won by the most highly evolved football culture.

So can we see a time when sporting strategy is entirely determined by data, not opinion? Will the argument be resolved by finding equipoise between evidence and intuition, facts and judgments? Or will empirical evidence gradually supplant myths and old wives' tales?

As captain of Middlesex in 2008, I used data from the Indian Premier League in our planning for the Twenty20 Cup. We won the competition. And yet, using "Moneyball" metrics, England would never have nurtured Andrew Flintoff into one of the world's best all-rounders. It was raw potential, not numbers, that led his coaches to invest so much faith in him. Even in baseball, teams using "Moneyball" methods have a mixed record.

Re-reading "Moneyball" now is a surprising experience. I found myself wondering if sport might go too far towards ultra-rationality, as it once, without doubt, leant too hard on untested hunches and folk wisdom. Lewis often compares sport's quants to Wall Street's derivatives traders, and means it as a compliment. These days, in the aftermath of the financial crisis, it is harder to accept that maths can make human judgments obsolete. From one perspective, "Moneyball" was a brave new dawn. From another, it looks like the high watermark of millennial over-confidence.

We will carry on picking and choosing our principles to suit our convictions. I am intuitive, you are strangely convinced, he is delusional. Or, if you prefer, I am rational, you are a reductionist, he is a slave to numbers.

Photograph Action Images