Bits of Books - Books by Title
The Numbers Game:
Everything You know About Football Is Wrong
Chris Anderson and David Sally
It is the book that could change the game for ever.
Football has long been waiting its Moneyball moment. Over the course of the past decade, the sport has been deluged with numbers. Clubs have spent millions establishing their own analysis departments and companies such as Opta and Prozone have gathered billions of data points, telling us how far players run, how many times they pass — anything and everything, in the hope of unlocking the game’s secrets, of finding a recipe for winning.
The problem is that nobody quite knows how to make sense of that avalanche of information. Does everything that can be counted need to be counted? What does all of this mean?
Chris Anderson and David Sally hope to be able to shed light on that. In The Numbers Game, they sift through that information to work out how clubs are using all of that data, and ask quite what it might mean for football’s future. The ramifications could be profound, marking a sea change not just in what we think we know about the game, but — as shown here, in this exclusive first extract from the book — how we think we should play it.
Is attack more important than defence: is it better to score, or not to concede?
In football, there is no more beautiful premise than the idea that attack always wins out. Millions of pounds ride on it every season, as teams clamber over each other to sign world-class forwards, paying them ever more eye-watering salaries. Have the best attack, and no defence will be able to stop you on your quest for glory. So goes the rough logic.
The question, though, is whether this approach is actually correct. Is one really more valuable than the other? Are we playing the game the right way? We collected 20 years’ worth of results across the top four European leagues. The first question we asked was this: do clubs that score the most goals always win the league? The simple answer is no. On average, clubs that scored the most goals in a season won only about half (51 per cent) of the championships available.
So do the clubs that concede the fewest goals win titles? Again, not necessarily. Our figures show that the best defence will pick up a championship 46 per cent of the time. Scoring the most goals over the course of a season gives you slightly better odds of winning the league than conceding the fewest, but as a strategy for guaranteeing a championship it seems to fall way short.
Winning a title, of course, can come down to the finest of margins — witness Manchester City’s last-gasp Premier League victory in 2012 — so this is far from a conclusive test. A better approach may be to see whether league position is more strongly associated with goals scored or goals conceded.
So here on the next chart are ten years’ worth of Premier League data (from 2001-02 to 2010-11) of goals (scored and conceded) and points won by the clubs that competed across the seasons.
They show that there are two strategies for producing points in the Premier League: you get more points if you score more goals, but conceding fewer is equally effective. The steepness of our trend lines is similar and both sets of points cluster tightly to those lines. These numbers suggest there may be more to the story than siding with attack over defence or vice versa. Maybe football is a sport of shades of grey.
Of course, this does not show us how teams won these points. They could have done it by winning games, or they could have done it by avoiding defeat. Which of those is more effective? Is it better to win, or not to lose?
We conducted a set of rigorous, sophisticated regression analyses on our Premier League data. Between 2001-02 and 2010-11, our data shows that scoring ten more goals was worth an additional 2.3 wins, while conceding ten fewer goals was worth 2.16 additional wins in the Premier League. What does that mean? Goals created and goals prevented contribute about equally to manufacturing wins in English football.
It is when we look at the number of games a club could expect to lose, though, that goals scored and goals conceded begin to vary in significance. Scoring an additional ten goals reduced a club’s expected number of defeats per season by 1.76; conceding ten fewer goals to our opponents reduced defeats in the Premier League by 2.35 matches. So when it came to avoiding defeat — getting something from a game — the goals that clubs didn’t concede were each 33 per cent more valuable than the goals they scored.
Does possession matter?
There is a philosophical tension within football, between those who prefer to see the ball swept around the pitch in beautiful patterns, the game played by Barcelona and Arsenal and Spain, and those, José Mourinho and Sam Allardyce and the rest, who prefer to see attacks carried out quickly, efficiently and devastatingly.
Thanks to the success of Barcelona and Spain, passing is in fashion at the start of the 21st century. Possession, the theory goes, helps you win games. Have more possession, win more games.
We are not concerned with theory. We are concerned with facts. We wanted to know whether keeping the ball better gives you an improved chance of success. If possession matters, after all, we should see it reflected in results on the pitch.
We looked at 1,140 matches over three Premier League seasons. That is 2,280 team performances. The answers we found were clear.
In attack, teams who do a better job of keeping the ball do have more shots and do score more goals. In defence, they restrict their opponents to fewer attempts and they concede fewer goals. They have more shots on goal and suffer fewer against: teams that pass the ball well outscore their opponents by 1.44 to 1.19 goals per game, and they outperform them by an almost identical margin defensively. The data also shows that, whatever statistic you look at — overall number of passes, completion percentage — possession of the ball increases offensive output.
When we looked at possession in another way, in terms of how rarely a team gave the ball away, we found equally important effects. Teams that turned the ball over less than the other side outscored their opponents by roughly 1.5 goals to 1.1; they outperformed them defensively by a similar margin. Keeping possession of the ball helped teams score more goals and concede less by about 0.3 to 0.5 goals at both ends of the pitch. That’s almost a goal per match. The maths is pretty simple, in the end.
It seems a fairly natural conclusion, then, to assume that more possession should lead to more wins and fewer losses. And it’s quite right, too: keeping hold of the ball, completing passes at a higher rate and not surrendering it to the opposition as much means more wins, more points and more success. Teams that had the greater share of possession won 39.4 per cent of their games, compared with just 31.6 per cent if they had less. In fact, however possession is measured — volume, completion, or overall — having more of the ball generated between 7.7 per cent or 11.7 per cent more wins.
Avoiding turnovers, though, is the most potent weapon of them all. The teams that had less than half the turnovers in any given match won about 44 per cent of the time, while those that gave the ball away more won slightly less than 27 per cent of the games. Having the ball is good. But not giving it back is better.
Do corners work?
“How many countries can you think of where a corner kick is treated with the same applause as a goal?” Mourinho memorably asked. “One. It only happens in England.”
He is quite right: corners, in the Premier League and the Football League, are seen as almost the next big thing to a goal. And why not? Just watch the procession of goals from set-pieces on Match of the Day and you have proof they are reliably profitable. Aren’t they?
Well, no, they aren’t. With the help of StatDNA, we examined what happened after a corner in a sample of 134 Premier League games from the 2010-11 season, a total of 1,434 corners. We were expecting proof that corners lead to shots and shots lead to goals. We expected a degree of slippage — not every corner leads to a shot — so the success rate of corners leading to shots is unlikely to be 100 per cent.
What we did not expect was that just 20.5 per cent did. Only one in five corners leads to a shot on goal. There is yet more slippage when we look at how many of these shots created from corners lead to goals. Here, we see that one in every nine shots produced from corners leads to one team celebrating. Put another way, 89 per cent of shots created by corners are wasted. How does that translate into real terms? The average Premier League team scores a goal from a corner once every ten games.
Is English football different?
We are told that in Italy, football is more defensive, in Spain it is more elegant and in the Premier League more physical, more exciting. Football’s culture changes from country to country and continent to continent. We all know that.
There is an abundance of data to support that idea, of course, from formations — figures from Opta Sports show that Spanish clubs used a 4-2-3-1 formation in 57.8 per cent of all matches they played during the 2010-11 season, while English teams did so in only nine per cent of theirs — to discipline. While the average Premier League match saw 24 fouls in 2010-11, referees in La Liga whistled for a foul 34 times per match.
But none of these minor discrepancies affects the outcomes of matches. They are incredibly similar across the top level of football in the 21st century. The most essential elements of the sport differ very little across countries and leagues, and in particular, the rare, the precious goal.
In fact, so sure are we that football looks the same among its ultimate elite, we have prepared an experiment of the nature of scoring in the very best leagues, the Bundesliga, the Premier League, La Liga and Serie A.
So below is a chart that shows the goals that were scored in the average match in these leagues in the ten seasons from 2001. We have declined, though, to distinguish between which graph pertains to which league.
Can you tell which one is which? If not, don’t feel too bad. We can’t either. It does not matter where you are playing or where your players are from; the essential features of the game, goal production and goal prevention are as similar as you can imagine.
But what about how those goals are produced? Here, too, other features of the game that we can count — and that are worth counting — things such as passes and shots, look very similar. Indeed, Opta figures from the 2010-11 season show that the average team, in the average match in the top four leagues in Europe contained between 425 (Bundesliga) and 449 (Serie A) passes. In Italy, only 54 of those passes were long, whereas the high was in Germany, too, with 59; those two countries provided the bookends for short passes, too: Germany with 332 per game, Italy 356. The differences between nations are cosmetic, shallow. That pattern holds true for everything else. The data also shows that teams took roughly the same number of total shots on goal (14) and shots on target (4.7), they earned a similar number of corners (about five) and they are awarded roughly the same number of penalty kicks (0.14) per game. We also found that that the number of free kicks, crosses from open play or headed goals were much the same.
The game is the same across the world’s elite football leagues. If it was not for the shirts, you would not be able to tell them apart.
Can a team be lucky?
Martin Lames, a dapper man in his early fifties, is Professor of Training Science and Computer Science at the Technical University in Munich. That may not sound that exciting. In reality, it means he watches football for a living, all in the name of science.
Lames has spent years developing computer and coding systems that allow researchers like him to record and analyse what happens on the football pitch and why. One of his favourite topics is luck.
Along with a team of collaborators, Lames has used his technology to record instances of good and bad luck on the pitch. Goals in particular are tailor-made for this kind of analysis: some are clearly produced intentionally, the result of hard work on the training pitch or superhuman vision from a wonderfully gifted player, and others are a little bit more fortunate: an unplanned deflection, a spilt cross, a missed tackle.
To assess how much luck plays a role in scoring goals, Lames and his match-watching, goal-scrutinising collaborators defined shooters’ good fortune as resulting from one of a total of six goal situations where shooters did mean to score, but the goals they did score had a strong and detectable element of being “unplanned” or “uncontrolled”.
Lames and his team have watched videos of more than 2,500 goals over the years, carefully coding each one for instances of luck. And so, after all those hours and hours of watching goals, how many did Lames and his team qualify as fortunate, as owing more than a little to luck? The figure was 44.4 per cent, though that varied a little from league to league and competition to competition. Lucky goals are particularly common when the score is 0-0. “That is when teams are still playing according to their system,” says Lames. “Something coincidental has to happen for a goal to be scored.”
Those 44.4 per cent of goals contain a detectable, visible portion of good fortune. As Lames’s work and our research in The Numbers Game show, football is a 50-50 proposition; well, almost. The game you see this weekend, the one that leaves you in a state of utter jubilation or bitter disbelief is really just the flip of a coin.
Tomorrow, in our second extract, we turn our attention to the managers: do they matter, how can we judge who is best and — as Chelsea, Manchester City and a host of others consider whether to fire the man currently in charge — does sacking them really work?
More books on Sport
Books by Title
Books by Author
Books by Topic
Bits of Books To Impress