Authors: Michael Calvin
Benham is a close friend of NBA legend Steve Nash, the LA Lakers point guard. He senses the significance of Battier’s admission that ‘when there’s any question, I trust the numbers. The numbers don’t lie.’ Benham has accumulated a substantial personal fortune at the age of 43 because of his nerve, and faith in the nuances of arithmetical logic. It has enabled him to secure Brentford’s future, by buying the land for the new stadium they hope to build by 2016–17 season, and given him a healthy scepticism about football’s innate conservatism:
‘I am quite alone in this belief, to say the least, but Shane Battier and the plus–minus principle is the way forward for football. Basketball informs us players’ individual stats don’t tell you much. The whole thing about Shane was that they looked at him in relation to the team. To define that in football terms, I am not just talking about whether a team wins or loses or scores or not, because there is a huge amount of what we call noise in that statistic. I want to look at the number and quality of the chances they created.
‘If I am looking at a striker I absolutely do not care about his goalscoring record. For me, the only thing that is interesting is how the team do collectively, offensively and defensively, within the context of an individual’s performance. To give an example, back in the day, I always thought Alan Shearer gave his team an amazing outlet. He’d be up there on his own. The team would be under pressure. You knew if you hoofed it in his general direction there’d be a high chance that the ball would stick, and he’d win a foul. That is a great defensive service Shearer is offering. The fact that a front player can hold the ball up means you are not going to concede another attack.
‘We’re working towards a new football model, but we are pretty far away. It really is not straightforward. It is a computational issue. Say you’ve got a Barcelona reserve, Tello for instance. To appreciate his value and potential you need to identify the players he has played with. Has he come on when Barça are 3–0 up and holding on to the ball? What is the context? Who were they playing? Similarly, you would look, for instance, at what happens on the few occasions Messi is absent. What impact does that have on Xavi and Iniesta?
‘I know people’s brains will fuse. They will tell us we are talking absolute bollocks. We are almost operating in a parallel universe. The thing about bogey teams is bollocks. If I tell that to anyone who has played the game they will tell me it is because I haven’t done so. I don’t dispute football people believe in the concept, but it doesn’t stand scrutiny. Many things which people think are important are not. Let’s take this idea that a player is “on fire”. If I choose a player at random it doesn’t matter if he has or hasn’t scored in ten games. It makes him no more or less likely to score in the upcoming game. People perceive confidence, or otherwise, where we simply do not see it.
‘If you look at any striker, at a random point in his career, and you want to know whether he is going to score in the next game, knowing how many he has scored in the last ten, twenty, forty games tells you very little. Similarly, the notion of “Hot Hands” in basketball is a myth. They used to think that a player with hot hands, one on a streak, was more likely to sink baskets. But then they did a bigger study and realised it wasn’t true. This didn’t go down well with certain people.
‘They tried to explain it away by saying that a hot player would take more risky shots, or be marked more tightly. Nonsense. It is the same with free throws. It doesn’t matter if the guy is hot as fire or cold as ice. If you have a guy like Steve Nash, who has taken several thousand free throws in his life, and you want to know what the percentage is he will get the next one, you need to know his percentages over the last five or ten years. If you only look at how he has done in the last five or ten games there is a huge amount of noise.’
An observation by Billy Beane, the
Moneyball
pioneer, sprang to mind: ‘The idea you would not rob good ideas from other sports and apply them is crazy. You would have a hard time finding any major sport in the world which is not using metrics in some way. Performances in baseball are much more easily measured than in soccer but each sport has a metric which is relevant. Identifying it is the trick. Basketball is much more similar to soccer and many NBA teams are using metrics.’
Benham is an intriguing, reclusive character. He is softly spoken for someone so dependent on, and comfortable with, the strength of his convictions. Few Brentford fans would recognise him, though his formative gesture, bunking off school to catch a Football Special to a fourth round League Cup tie at Nottingham Forest in December 1982, has entered club mythology. Like any supporter, he relishes landmarks, such as winning the old Third Division at Peterborough in 1992, and such eccentric achievements as coming back from 4–1 down, with ten men, to draw 4–4 at Doncaster in 1982–83. The idiosyncrasies and inadequacies of any small club mask the potential which his wealth has unlocked.
When Brentford resorted to taking bucket collections to ease cashflow problems, he took on outstanding loans approaching £10 million. Within ten days of assuming ownership, in June 2012, he ended eight years of procrastination by buying the site for the new Community Stadium in Lionel Road, less than half a mile from Griffin Park. He is anxious not to be bracketed with the tangerine-hued self-publicists, vulpine opportunists and imported control freaks who inveigle their way into club ownership, and his allegiance to Brentford is genuine. He funded the upgrade in the club’s academy, despite his understandable reluctance to trust the Premier League and FA to govern youth football equitably.
We conducted this, his first major interview, in a small, glass-walled meeting room set on a landing above the trading floor at Smartodds, on the third floor of an unprepossessing industrial complex close to Kentish Town tube station in North London. His visibility to his staff, uniformly responsive and largely employed straight from university, was indicative of an open management style in what is, by its nature, a secretive environment. He padded around the office, where lines of video monitors and computer terminals played out an array of live matches from around the world, in white towelling socks, combat trousers and a lightweight hoodie.
Football is the primary focus, although Benham also works in basketball, baseball, American football, ice hockey and tennis. It is an arcane world of parameters, probability theory and pseudo likelihood, which has a human dimension. Benham orders each recruit to read
Thinking Fast and Slow
, a book by Nobel prize-winning psychologist Daniel Kahneman, a professor emeritus at Princeton. Two chapters, ‘The Illusion of Validity’ and ‘Intuition Versus Formula’, are deemed particularly relevant:
‘In the fifties, Kahneman was helping the Israeli Army on a leadership exercise. Six trainee officers, in the blazing sun, had to get a log over a wall. They and the log couldn’t touch the wall. Very difficult, but you should be able to spot leadership potential. He was looking to see who was calm under pressure, who was respected by his peers, who was not respected, and who was flustered. He then made predictions on who would make officer material.
‘It turned out his predictions were terrible, no better than chance. He was astonished because he had huge certainty that his predictions would be good enough. They were absolutely useless. What was especially interesting was that when he watched another batch of people do exactly the same thing, he couldn’t help but feel incredibly confident in his predictions, despite previous experience. He concludes that experts in so many fields are ludicrously overconfident in their own intuitive ability to make predictions.
‘That’s why I make all our traders read those two chapters again and again. We have very sophisticated mathematical models, rating teams. What was constantly happening was that our traders, like everyone else, thought of themselves as football experts. I’ve done the same thing, and given my muggish public judgements much more weighting than a model that’s been put together by loads of Maths PhDs and extensively back-tested.
‘It is a phenomenon I see again and again in football. If I want to know how good a player is I want to speak to a person who has seen that player play one hundred times, in all conditions. What tends to happen is so many people in football will see that player for forty minutes and decide they are the oracle. They absolutely know. A typical example involved Gerard Houllier, when he signed the United States midfield player Michael Bradley from Borussia Mönchengladbach for Aston Villa.
‘He was asked whether he had scouted him, and replied he had watched him in the World Cup. He said “he played four games. That was enough for me.” Four games is a tiny, tiny sample. Anyone can play particularly well or badly in that timeframe. Perhaps most importantly, those four ‘live’ games in South Africa were for the United States, against England, Slovenia, Algeria and Ghana. England are OK, but the others are Championship-level teams.
‘What tends to happen with human beings is once they form an opinion, it is very hard for them to shake that opinion. The best player we have ever had at Brentford was Wojciech Szczęsny, on loan from Arsenal. He was outstanding but in one game towards the end of his loan spell, against MK Dons, he dropped two and was responsible for two goals. Anyone seeing him in that game would have said dodgy keeper, flapper, but we knew otherwise. You might see a full back have a horrific game, but what are the variables? Is he playing injured? Does he have personal issues? Is it not his fault? Is the midfielder in front of him not giving him cover?’
Football engages and elevates; it bestows bogus authority and excuses crass ignorance. The average crowd is swayed by the power of perception and prejudice. Loyalists look for the unjustified alibi, while cynics cherish a glass that is perpetually half empty. Throw in regurgitated clichés from the golf club bores on
Match of the Day
, and it becomes clear why scouts, coaches, managers, players and even professional gamblers are routinely second-guessed. Benham smiles at the eccentricity of the court of public opinion:
‘It’s a funny one. Fans, and the media, see something in everything, when it is actually randomness. For example, if a team goes on a lucky or unlucky run, there is nothing really more to say than they have been very lucky, or otherwise. People feel obliged to say the lucky team ground out a result. They embellish the story by saying they showed character and commitment. The old lines – “the sign of a good team is when you win when not playing well” – are trotted out. It is all bollocks.
‘Within every single move in football there are hundreds of small actions, little pieces of luck and skill, good and bad play. In professional gambling you try to strip out the element of luck. You are left with the signal after you have got rid of the noise. The more low scoring the sport, the more randomness there is. If the better team always won in sport then no one would watch it. There needs to be some element of randomness but not much. I’ll quite often watch NFL games when one team wins 35–0 and they are the inferior team. Maybe the other team had double the yardage, but couldn’t convert.’
As the season evolved, Damien Comolli surfaced intermittently to justify his legacy at Liverpool. Charlie Adam had been sold to Stoke, and Andy Carroll was enduring an injury-interrupted loan spell at West Ham. Yet, to be fair to the Frenchman, Stewart Downing had revived his career after being informed he was surplus to requirements. Jordan Henderson was maturing, appreciably, in a team energised by the pace and vivacity of Raheem Sterling.
Comolli’s rumoured interest in succeeding Dan Ashworth as sporting and technical director at West Bromwich Albion remained unrequited, but his profile caused Benham to pause. He had not forgotten a clandestine meeting with Comolli, orchestrated by Frank McParland, whose stock as Liverpool’s academy director had risen as Sterling had become an England international, and the most coveted young player in the Premier League:
‘Is a statistical model useful in scouting and recruitment? Absolutely, without a doubt. The problem, however, is that too many people in football are running these computer models without having a maths or science background. If we had a hundred of the best maths PhDs we could find and they were working for the next ten years, there would still be loads of uncertainty. We’d probably have five different models. Ask them to spit out a number on a particular game, and they might well come out with wildly different numbers.
‘The thing about modelling is that it is an inexact science. You are constantly running into lack of data versus relevance. If I want to model home advantage in the Premier League for instance, and I only look at one year, I don’t have enough data. If I look at the last hundred years I have loads of data but little is relevant. There is a constant offset involved. For instance, do I look at other top leagues in Europe, or lower leagues in England?
‘Frank took me up to Liverpool to meet Comolli. The idea was “we’ve got a model, you’ve got a model, maybe we could talk about ways in which we could work together”. It was extraordinary. He just said, “my model is correct.” Anyone with a pure maths or science background simply wouldn’t have said that. Science, in particular, is all about uncertainty and the limits of human knowledge. He was laughably overconfident.
‘Looked at logically, it is easy to see how a club could end up paying £35 million for Andy Carroll. That would need reliance on a model which didn’t correspond to the reality. In those circumstances there is a temptation to tweak the parameters, mix them up, until it starts to get sensible answers. Carroll had an exceptional six months in the Premier League with Newcastle before Liverpool signed him. A model might tell you he won a high number of aerial duels but only over a handful of games. We have to accept there might be some noise in that.
‘Valuing new players is always a tricky one. A lot of models used by clubs seem to create what I call over-parameterisation. They have far too much in them. As I said, if I am looking for a striker I am not interested in his scoring record at all. All that is important to me is the team creates more chances, and offers fewer, when he is on the pitch. If there is a gun to my head, and I really have to include a scoring record in the model, it would be as long term as possible. That means his record over five or ten years.