in Robin Bennett by Robin Bennett. Permalink.

Robin Bennett visits extraordinary people who have made money in unusual ways. Like Steve, whose genius led him to write an algorithm that allows him to exploit a dangerous industry: gambling.

Steve looks like Sting got up one morning and decided to put all his clothes on inside out.

Scruffy.

But then again, he can afford to be. In the seven years I've known Steve he has gone from being a humble IT manager for Wallingford Town Council to never needing to work in his lifetime ever again. And he’s done it in just the way your granny warned you would lead to ruin and misery: gambling.

When I first met Steve he had developed a sort of party trick whereby, regardless of the outcome of any Formula 1 race, he would pocket around £500. Essentially, he was doing what other people had done before in the city trading futures, namely working out an algorithm that would turn the odds minutely in his favour when betting on the outcome of a multiple field.

The other essential element to the efficiency of the maths was that he was betting against other punters. The odds are therefore even, whereas, if he was gambling against what they call "the house" in Monte Carlo and Las Vegas, he would immediately start at a statistical disadvantage.

What made all this possible was the existence of websites where people gamble against each other.

Steve set to work on an algorithm that improved his better performance on sports with a constantly changing field (such as horse racing, Formula 1, golf etc.) as opposed to sports like tennis or cricket, where there is just a progression through a tournament, knock-outs and a winner.

“There are two elements to a bet: stake and price. Betfair offers the facility to set your own price and stake, unlike normal bookmakers where the price is fixed and the stake usually limited," Steve told me. "I a submitted bet can't be matched with an opposing one, then it's added to the website as an offer. The offers are arranged in order of price and the best three are displayed."

When placing new bets Steve would look at the price and stake of the first three entries and make a judgement as to where in the queue to place the bet, determined by the price he was prepared to offer. His stake would be determined by the amount he had decided to allocate to the market, divided by the price.

Essentially, the better price you offer the more likely you are to get the bet matched, but the smaller your edge. What you want to achieve is getting the bet in as far back from the front of the queue without reducing the chance of it being matched too much.

"This, I imagined at the time, was done with skill and judgement that would be impossible to program successfully; but that was just my ego trying to make out I was in some way important to the process."

"What I did, as with most computer solutions, was a bit dull and tedious. I simply went through numerous scenarios and, one by one, forced myself to justify my decision in quantative terms that could be translated into code. Each time a case deviated from the code, I updated the code."

When Steve had enough of this he simply - but nervously - ran it, albeit in a controlled way which turned the analysis on its head. Rather than him deciding where to place each bet, Steve would run the code and see where the program placed it.

"If I agreed I would congratulate myself and if I disagreed I would tweak the code again. In fact, what I ended up doing was thinking about the bet placement in a lot more detail than I ever would have done otherwise. The result was a program that was able to apply my whole thought process to every single bet it placed and I had to admit towards the end, much to my disgust, that it was better than me."

"I would look at its bet placement and it would sometimes take me a few minutes to figure out why it was right; but usually I would grudgingly agree that it was."

The system isn't cleverer than Steve, of course, but it's blindingly quick and never lazy or distracted. It can multi-task, playing many markets simultaneously, each market and each bet getting the best, focused decision-making applied to it.

Whilst this is only a small part of the overall program, it's a pivotal part and quite possibly the most "intelligent" part. Interestingly, much like the development of the code itself it's only by sitting down and trying to explain it to someone that Steve discovered (or maybe it's just remembered?) how he did it.

"You see why I can't write, I get bored and silly, but it made me laugh. So I left it underscored. Actually, I did take one bit out. My code is the same, full of expletives and childish references to people who will never see it."

It doesn’t always work. So, to an extent you still need a gambler’s nerve. On holiday last year Steve mentioned one of his best days - a win over a few tens of thousands -had been regularly topped by worse days - less tens of thousands, which can’t feel good.

Other people are doing this too, some on quite a grand scale where they are employing rooms full of people doing what he does. This all seems like hard work for Steve; and I tend to agree. I love the fact that he can monitor progress of bets placed on his mobile phone from anywhere in the world and simply make the necessary adjustments to the algorithm if profits take a dip before going back to his pina colada.

The feeling I have with Steve is that the hardest thing for him was taking the plunge; and he confirmed this when I talked to him.

“The biggest risk I ever took was the £200 a month fee I agreed to pay to Betfair as a subscription.”

Steve has benefitted hugely in a very short space of time from working out an advantage in an existing system and then having the guts to exploit it. For that he richly deserves everything he has got and universal accolade in spite of his dress sense.

Robin Bennett has started eight separate companies since 1992, each for less than £5000. He's the author of Start-up Smart - How to start and build a successful business on a budget