Rob Mastrodomenico uses data to estimate the outcomes of sports events for professional punters, and it's a complicated business
Rob Mastrodomenico has just got back from a lunchtime kickabout with his colleagues. "I'm still a bit out of breath," he says apologetically, despite looking considerably less out of breath than I do after the short walk from Kentish Town tube station. I assume they've all been playing on nearby Hampstead Heath but his reply is unexpected: "No, in a client's back garden."
Come again? "One of our clients has a massive back garden with a modified tennis court," he explains. "We see him round the office a lot and he invites us down to play. He enjoys a game as well."
Walking into Smartodds, where Mastrodomenico is a quantitative analyst ? mercifully abbreviated to a less tongue-twisting "quant" ? is a disorientating experience. It's a typically trendy, if functional, warehouse space but you wouldn't immediately guess its clients could be so well-heeled.
Only a clever poster by the entrance, showing a complex mathematical equation scrawled in chalk on a football tactics blackboard, offers a hint to the kind of work that goes on here. The firm is a consultancy producing statistical information about sports events ? mainly football matches ? for professional gamblers, who then use it to inform their betting patterns.
"A future game is a chance event, so we're trying to use old data to predict the outcome of those events as accurately as possible," he explains. "We're not trying to say this is going to happen, we're trying to say this will happen with a certain probability. If our probabilities are better than those of the bookmakers, then, in the long run, our clients will win money."
Judging by the back gardens kept by some of the firm's clients, this can be very valuable information indeed.
The other confounding thing is Mastrodomenico himself. Tanned, relaxed and wearing a blue T-shirt and khaki shorts, he seems more likely to have stepped out of rehearsals with a boy band than to have spent the morning contemplating dense mathematical permutations.
Watching him pad around the company's stylishly appointed warehouse offices, past the banks of software developers and quants gazing into multiple computer screens and finishing off their sandwiches, I start to wonder how difficult all this can actually be. Then I'm stopped in my tracks by a 10ft-long whiteboard covered with impenetrable algebra and the occasional doodle of a space rocket. "We don't really understand it either," he quips.
Clearly, there is more to this game than meets the eye. The company maintains statistical databases for every football club in the land, but to make any sense of the data, it must be run through a sophisticated computer model. The idea is all based on a research model into football outcomes known as Dixon-Coles (PDF), a co-author of which, Stuart Coles, I find lurking in a corner. "He wrote the seminal paper," explains Mastrodomenico, with a touching reverence. "It set the whole thing going. He'll be very modest about it, though." On cue, Coles smiles shyly.
Yet just as coverage of live sport develops, with multiple camera angles and videographic analysis, so, too, does the richness of the data that can be gathered, meaning models require constant fine tuning, and it's in this area that most of Mastrodomenico's attention is focused.
Added to that are the intangibles that pure mathematics simply cannot account for. A good example, he says, would be games between two very good teams, which often turn out to be tense, low-scoring encounters.
"The model estimates parameters for attack and defence for every team, and how they might affect one another. But with two very good teams, the model might not pick that up, and you might find it to be overestimating," he goes on. "So we just try to improve it. You're always trying to look and do something different to what you've got, to improve it."
A sizeable chunk of Smartodds's floor space is taken up with an area like the bridge of a ship lined with 70 or 80 television screens, each showing a different sports channel from somewhere around the world. To keep track of it all on busy nights, teams of "watchers" come in to monitor specific games on behalf of clients. "We're watching for certain things that maybe don't come up from regular stats," he says, being careful not to over-elaborate. Like what? I ask. But he just smiles cryptically.
If they wish, clients can come in to use the facilities and statistical analysts can be on hand to provide further breakdowns if needed. "One common example is that pundits often say that if you play in the Champions' League on Wednesday, you might play badly in the Premier League on Saturday. So we might go away and study that, then we can tell our clients whether it's actually true or false." And is it? Again he grins politely.
Despite having been at Smartodds for five years, Mastrodomenico ? a Swindon Town supporter with a broad Wiltshire twang to prove it ? still wears the mildly bemused expression of someone who can't quite figure out how his love of mathematics and football came to such a perfectly intertwined conclusion.
"For many of us here it's just a dream job," he says. "When we're not talking about maths, we're probably talking about football."
That's easy enough to believe judging from a quick look around the office, which is populated mostly by males of a similar age and demeanour. They seem a close-knit group and he agrees.
"You don't have to like sport to work here," he says. "Good statisticians are good statisticians. People come in because they like it. If everyone was really dour it'd be terrible. You've got people who really enjoy the modelling side in our team, then you've got the analysts who just love chatting about football."
In fact, his role at Smartodds requires a surprisingly wide variety of different skills. "In our team it's very hands-on," he says. "We do a lot of statistics but there's also a lot of [computer] coding. You might have to explain technical things to clients as well, so you vary it up a bit."
These are not the only differences between his work and that of more traditional statisticians, who might typically have a fixed data set and then set out to build a model that will fit. "We're building models to predict things that will happen in the future," Mastrodomenico points out.
His personal chances of achieving such a harmonious career outcome didn't seem quite so high when, having completed a degree in mathematics and statistics at the University of Reading, he drifted into a genetics-based PhD.
"I'm the exception, not the rule," he admits. "I was going to do a master's, which you generally need to do any decent statistics job. I got accepted on to the MSc at Reading, but because I knew some people there, I got offered the PhD without really knowing what I was letting myself in for."
Despite enjoying the research, he was uncertain as to what to do after completing his doctorate. But an advertisement for football modellers on the mailing list Allstats led him to Smartodds ? then a fledgling start-up ? where he has stayed ever since.
"When I joined, there were three people on my team. Now there are 18. I'd say every team has grown by at least a factor of four," he says.
Yet if his story makes the route into this line of work sound straightforward, it is anything but. "Most of the guys there have PhDs, if not studying towards them," he says, motioning to the desk of quants behind us. "We ideally look to get people with research experience and there's also people who already do what we're doing, so it's quite competitive. You'd have to have at least read Dixon-Coles and understand a bit about what we're doing."
With the football season now over, he has time to turn his attention to other projects, including some of the models the company is developing for sports in the US market, such as American football, hockey and basketball. It's a challenge he regards as a pleasure rather than a chore.
"A lot of the guys mess around with models in their spare time, just for fun, because if you know how to build one, it's quite interesting just to play about with it," he points out. "I know that sounds a bit nerdy but, you know, we are a bit nerdy."
But it's time for him to get back to his models. Before leaving I check on the arrangements for the photographer's visit, and Mastrodomenico tells me he's had his hair cut especially.
"I know I'm looking a bit laid-back, but I'm going to make a bit more of an effort," he says proudly. "I'm going to put on a smarter T-shirt."
Curriculum vitae
Pay A quant of Mastrodomenico's experience earns �45,000-�60,000.
Hours Essentially a 40-hour week, "not because you have to, more because you want to. With a lot of research you take it to the end. There's not a deadline but you're expected to say when you're done."
Work-life balance Mastrodomenico commutes from Swindon, an hour or so away by train. "People work to fit their lifestyle here, not the other way round. It's quite easy to work from home, but I think on research-type projects it's better to work with the people most of the time."
Best thing "It's all pretty good. You're doing statistics in a sports environment ? for a lot of people it's a dream job, basically. In terms of statistics you're the prettiest girl at the ball."
Worst thing "There's nothing that bad. Personally, it's the commute. There's frustration, sometimes, when you're working on something new and don't know where you're going, but that's typical research frustration."
Overtime
The biggest myth about Rob's job 'is that we're all degenerate gamblers, when in fact we don't bet at all. We're privy to some very sensitive commercial information and we don't want to blur the lines. So we steer clear.' For lunch Rob has 'soup, Monday to Thursday. Pretty boring. On Friday I'll treat myself to a sandwich.' Rob plays guitar in an indie rock band. They used to play a lot of gigs, but at 30, he feels his ambitions of being a rock star might be on the wane. Rob's favourite band is "the Foo Fighters. Or of all-time, the Beatles".
Source: http://www.guardian.co.uk/money/2011/jun/11/working-life-quantitative-analyst
Petrofac Manchester United War crimes China Winter sun Communities
No comments:
Post a Comment