The relationship between analytics and randomness in football

The relationship between analytics and randomness in football

For the second time in three years, the two biggest teams in Madrid faced each other in the Champions League final. Real Madrid had endured a somewhat disappointing season, unable to fight for the La Liga title and even going through a coaching change, so the game offered them a chance to sort of redeem themselves.

Atlético also failed to win the Spanish title, but somehow second place seems like an achievement for them, and reaching a second Champions League final in such a short period of time was undoubtedly a testament of Diego Simeone’s work at the helm of Los Colchoneros. Although they had lost to Real two years ago in Lisbon, a run of good results against their neighbouring rivals must have led them to believe they could leave Milan – where the final took place – in a happier mood.

Things, however, didn’t go as well as they hoped. Real took control of the game from the outset, creating opportunities and busying Atlético’s goalkeeper Oblak – as well as the team’s supporters. By the 15th minute, Sergio Ramos turned one of those opportunities into an actual goal, striking the ball into the back of the net by way of a Toni Kroos free-kick and a Gareth Bale assist. On TV, the instant replays instantly made it apparent that Ramos was actually offside when Bale touched the ball, but the referee and his assistants failed to realise it, so the goal stood.

Atlético tried to turn things around, keeping the ball and looking for a goal of their own, but their efforts in the first half were mostly futile. Real were well organised defensively, covering up enough space so as to prevent their opponent from breaking in and threatening their goal.

Atlético came back from the break hoping that things would turn around in the second half, and for a brief moment they must have thought they’d got what they’d prayed for: Fernando Torres got himself into the Real area before he was brought to the ground by Pepe, giving Atlético a penalty kick. Charged with taking it, Antoine Griezmann could only shoot the ball towards the bar, wasting the chance to nullify Real’s advantage.

Atlético kept pushing though, as Godín, Savić, Koke, Saul and Torres all chanced their hand at the Real goal. But Real weren’t just parking the bus, instead trying to score a second goal and kill the game. The usual suspects – Bale, Ronaldo, and Benzema – had several chances, all meeting either the Atlético defence or their goalkeeper, but never the back of the net.

And then, in the 79th minute, Yannick Carrasco, who had started the game sitting on the bench but came into it as a substitute at the beginning of the second half, met a pass from Juanfran and sent the ball across the goal line for the equaliser. With just over 10 minutes to go, no-one would’ve blame either team for turning cautious and slowing the game down so as to save their strength for the likely extra 30 minutes of play. But neither of them did so. They pushed for another goal, trying to end things right then and there. Unfortunately, they were both unsuccessful, and the game went into those extra 30 minutes and then – after the score remained the same –the penalty shootout.

Lucas Vásquez scored the first for Real. Griezmann, this time, didn’t miss, and Atlético also converted their first attempt. Marcelo, Gabi, Bale, Saul and Pepe all scored for their respective teams, and so when Juanfran walked towards the area to take Atlético’s fourth attempt, Real were ahead 4-3.

The pressure was too much for him, and he missed. Ronaldo was set to take Real’s fifth attempt. He had a relatively discreet performance, but finally he had a chance to end things and give Real their 11th European Cup. He didn’t waste it, ripping his shirt off with joy after his decisive kick.

• • • •


Read  |  Diego Simeone’s defining time at Catania

• • • •

After the game, Simeone had resigned himself to the outcome: “We began the game badly, but got better after about 25 minutes,” he said. “We had a chance to equalise quickly in second half, but we did not take advantage of it. They had a chance for 2-0, but we equalised almost immediately afterwards. Then the game became very tactical, very draining, and the two teams were pretty tired.

“Today was not meant to be for us. There is no such thing as justice in football. Whoever wins deserves to win. There are no excuses. They were better than us again.”

If we consider Michael Caley’s expected goals calculations for the game, Simeone was right. Expected goals assumes that while the number of shots on goal a team can convert into an actual score varies significantly from game to game and is dependent of many factors outside of that team’s control, they reflect the quality of the team’s play.

Although in any given match each of the competing teams can be lucky or unlucky and score or be scored on – or fail to score or be saved from being scored on – because the ball bounced on a bad pitch and hit the foot in such a way that it went off the mark, or because the shooter’s teammate jumped in front of his shot and blocked him, as a season or tournament progresses, a team’s exposure – beneficial or otherwise – to such random factors will “regress to the mean” and so will their outcomes.

Looking at both the number of shots a team makes and concedes (itself a predictor of future performance far more reliable than a mere win/loss record) and the quality of those chances (taking into account the place from where it was taken, the type and location of the pass than originated it, the speed of the attacking play that led to it, and several other factors), “Expected Goals” allows us to see which team in a given game performed more consistently, or in such a way that is more indicative of that team’s “real” outcome without the distorting effect of randomness.

In the Champions League final, according to Caley, Real Madrid had, over the 90 minutes of regulation time, a goal expectation of 2.3 goals, against Atlético’s slightly lower 1.4. Real really were better, and like Simeone said, they did deserve to win.

But, unlike what Simeone said, Real didn’t really win. If this had been a “normal” game with only 90 minutes of regulation time and no extra-time or penalties, the two teams would have tied with each other. And considering that Real’s only goal was an illegal one, they may have been the better team in the 90 minutes they were supposed to play, but they might have even lost the game.

Simeone may have been right in saying that there is no justice in football, but not in the way he meant it: by their Expected Goals, Real would have been the just winner of the Champions League final, and Atlético managed to get an unjust draw.

Perhaps this is why so many football fans are skeptical of advanced analytics: they are a quantifiable proof that the winner of a game – or even of a whole season’s league – is not necessarily the best team, and that is an affront to the way we all tend to look at competitive sport.

The sole purpose – aside from the provision of mass entertainment – of a sports competition is the finding out of which competitor is the best. This is why a poor refereeing decision so enrages a team’s fan;: it’s almost a stab to the sort of cosmic, pure justice we expect to see reflected in sports. In the end we seem to believe, if all is fair and square, the best should win. Ultimately we are wrong. Especially when it comes to football.

Bill Barnwell, who writes about the other football for ESPN, often points out how there are countless random minute factors that can help push the odds in one team’s favour or another, how if you put two great teams on a neutral field a million times, even the one with superior match-ups isn’t going to win every each one of them.

• • • •

What if Griezmann had scored his penalty in normal time?

• • • •

A win-loss record might be a poor method for the appraisal of a team’s calibre, for, as Barnwell would note, “all wins aren’t created equal”. If that is true of pretty much every sport, it is particularly true of football, where – according to Chris Anderson and David Sally, the authors of The Numbers Game – the outcome of a game is 50 percent luck.

As John Tierney wrote in his New York Times column: “If you regard a soccer match as an experiment to determine which team is better, then it’s not much of an experiment. It involves hundreds of skillful moves and stratagems, yet each team averages only a dozen shots, and the outcome is decided by several quick and often random events. In most games, no more than three goals are scored, and the typical margin of victory is a single goal. To a scientist, the measurements are too few to draw a statistically reliable conclusion about which team is more skilled. The score may instead be the result of measurement error, a.k.a. luck.”

There are so many things that can impact a game that are beyond the participants’ control, from injuries to players to who wins the coin toss for the first attempt in a penalty shootout (according to Anderson and Sally, that team wins 61 percent of the time), that our idea of the purpose of a competitive game borders on meaningless.

Analytics like Expected Goals, by measuring a team’s performance independently of their win/loss record, point out to fans the futility of looking at football games and competitions as a reliable way to measure who is the best at playing the game. When a match – or a league title – can be decided by a ball being deflected by a beach ball that has fallen into the pitch or by a player slipping in the grass, it’s easy to see how a team like Leicester, who according to Paul Riley’s own Expected Goals model was only the fifth best team in this year’s Premier League, can come away with the title, while Tottenham and Arsenal consistently outperform their rivals and still fail to reach the end of the season at the top of the table.

To make matters worse, acknowledging the role of randomness in football forces us to also acknowledge its role in our lives. Both in watching football and in our everyday life, we tend to misjudge how heavy a role luck plays in the outcomes of our pursuits; we fail to acknowledge how so much of what happens to us is beyond our control; we fail to grasp the Sliding-Doorsyness of it all; how even something like finding a job, or the right person to spend your life with, is to a large degree a matter of being fortunate enough to be in the right place at the right time.

If, in 1984, a struggling actor from New York didn’t decide to go to LA to either watch the Olympics or visit a girlfriend (the accounts differ), he wouldn’t have been convinced by an agent to take the opportunity to go to a few auditions, and he wouldn’t have been chosen to play the male lead in a TV show called Moonlighting, and no one outside his family and circle of friends would have ever heard of the name Bruce Willis.

If in the 1970s a Spanish gentleman hadn’t made a mistake in multiplying seven times seven (he had “dreamed of the number 7 for seven straight nights”), he wouldn’t have bought a lottery ticket with a number ending in 48 (for, as he mistakenly said, “7 times 7 is 48”), and he wouldn’t have won the El Gordo and earned himself a fortune.

And if my own father didn’t have such an interest on the subject of randomness and a copy of Leonard Mlodinow’s The Drunkard’s Walk: How Randomness Rules Our Lives, I wouldn’t have been able to extract these two anecdotes and give some colour to this piece.

• • • •


Read  |  What is ‘good’ football? The role of aesthetics in the modern game

• • • •

Randomness is Mlodinow’s life. He is the son of two Holocaust survivors, his existence only made possible by the relatively unlikely circumstance of his progenitors having both outlasted Hitler’s lunacy. He grew up to earn himself a PhD in physics, and, having written for shows like MacGyver and Star Trek: The Next Generation,  now teaches about randomness at the California Institute of Technology. And his teachings will by now have a ring of familiarity to it for the reader of this piece: “A lot of what happens to us – success in our careers, in our investments, and in our life decisions, both minor and major – is as much a result of random factors as the result of skill, preparedness, and hard work.”

And yet, the randomness some football fans might not want to think about also might actually be the reason why they – we – keeping watching football. As Sally noted in an interview with Pacific Standard, “Not knowing the outcome is a big part of watching sports.”

Without randomness, the best teams – which, for financial and historical reasons, tend to be the same pretty much every year – would always win, and the outcomes of the games would be much more easily predictable. If football was a less random sport, it would also be less intriguing, less exciting, and probably less meaningful: Leicester might have been lucky in winning this year’s Premier League, outperforming their output in a way that likely is neither sustainable nor reproducible; but who, apart from Tottenham fans – and probably not even them – wasn’t even if only a little bit thrilled by the sight of the Leicester fans celebrating their historic achievement? Who didn’t find those images heartwarming? Who wasn’t thrilled to have been alive to see such an unlikely and surprising champion?

Randomness and its impact on every field of human endeavour may be a scary reality to face, but much like in sports, it is an intrinsic part of life. It’s natural to feel powerless if so much of what makes our lives good or bad, enjoyable or depressing, successful or unsuccessful, is beyond our control, so independent of our qualities or lack of them. However, we shouldn’t. For one, because randomness is not that bad of a thing.

As my father says, “Randomness was what brought us here.” It was a series of random mutations in our genetic makeup that equipped us human beings to be particularly fit to survive – and prosper – in our environment; we did nothing to “deserve” brains that allow us to speak with each other and create technological advances that better our lives, or opposable thumbs permitting us to utilise instruments; we randomly developed them, and because they were helpful in our endeavour to propagate our genes, they stuck around.

And secondly – to use Mlodinow’s phrasing –  although the outline of our lives is continuously coaxed in new directions by a variety of random events that determine our fate, they do so only along with our responses to them; we may have our circumstances limited by the power of randomness, but we have a say in how we handle them.

Leicester may have been lucky in winning the Premier League, but they were good enough to take advantage of the luck they were fortunate to have had. Atlético, on the other hand, may have been lucky in not losing to Real in the 90 minutes of the Champions League final, but they weren’t good enough to translate that luck into a win – not even taking advantage of being awarded a penalty kick in the second half. And when they needed to get lucky once again – in the coins toss before the penalty shootout – they ran out of luck.

Real, the team that deserved to win in the 90 minutes but wasn’t lucky enough to do so, was then good enough to take advantage of what randomness offered them in that coin toss.

Does that make sense? Or is it just a little too random for you?

By Bruno Alves. Follow @ba_lifeofbruno

No Comments Yet

Comments are closed