Software programs may offer more than an outside chance at the bookies. Mike Hewitt meets the gamblers who swear by their PCs
Professional gamblers never talk about getting lucky. Rather, they talk about shortening the odds and maintaining their percentages. They know exactly when to place money and, far more important, when not to. Here is where a computer helps. It can coolly analyse odds and form and make a rational judgement based on statistics. It will not get carried away if there is a so-called winning streak, nor will it become despondent if it starts losing - which explains why many punters are now turning to PCs in an attempt to give themselves more of an advantage.

Actor and former croupier Peter Cartwright has been using his "creative pauses" to write a bookie-beating (he hopes) program called Microform. I put an early alpha version to the test on last year's Derby Day. From outside the bookmakers, I phoned Cartwright and told him the race-meeting and the current odds. He keyed them into his computer and it whirred away. "Five pounds on Blue Grass Prince," he said, and put the phone down. I burrowed my way into the betting shop through the cigarette smoke and did the necessary: pounds 5 on Blue Grass Prince at odds of 11-2, to win.

Three minutes later, the field of 10 runners was off. Blue Grass Prince was stuck midway down the field, apparently disinclined to shift himself. Then he surged forward, accelerating smartly past all the other runners. At the winning post, he was a good one and a half lengths ahead of the field - with my money on his nose.

Twenty-five pounds of winnings, thank you very much - and another phone call. The next race was the Epsom Derby. "Microform says the winner will be Erhaab at 7-2," said Cartwright. "But it also says there's no point backing him. There's more than 11 runners in the field - it's a lottery."

In other words, there were so many horses jostling for space, possibly boxing each other in, that it would have been like trying to manoeuvre a trolley through Sainsbury's on a Saturday morning.

So I did not back him - which was a pity. Anyone who had a flutter on Derby Day 1994 may recall that Erhaab did indeed win.

Microform, which should be available as shareware at the beginning of next year's flat-racing season, works by analysing all the variables - previous form, weight, the going, the draw and so on - and assigning them values in order to calculate an overall score.

If, for example, a particular horse usually performs well carrying x number of pounds when the going is good, and those same conditions apply today, then he will be assigned a high score. If, on the other hand, he is carrying 8lb more than usual and today the going is firm, he will be assigned a correspondingly lower score.

Microform, like all similar racing programs, can be effective only if its data is fully comprehensive. Cartwright therefore spends most of his waking hours feeding in horseflesh statistics from previous years, and then tweaking the program's algorithms to predict winners. So, how is he doing?

"Last year, I was making an average 12.5 per cent return on my stake. That's not bad, considering a professional punter usually looks to make about 10 per cent profit. This year, though, I had a long losing streak. But bear in mind that there's no such animal as a 'sure thing'. Last year's prize-winning thoroughbred can be this year's Pedigree Chum. Now, however, I'm beginning to pick up again. Currently, my profits are averaging around 15 per cent. And it's better than putting my money in the building society."

Blackjack player and World Memory Champion Dominic O'Brien has also been doing well. At one point he was earning pounds 1,000 a day. However, his "benefactors" soon cottoned on, and now he has been barred from almost every casino in the UK and a good many in the US. O'Brien is a "card counter". This means he analyses the play of a deck, memorises which cards have been dealt, and then either ups or reduces his stake accordingly.

His strategies have to vary because different casinos employ different methods of dealing. Some might use four decks, for example, while another will use two. Also, if you study a specific shuffle carefully, based on previous analyses of similar shuffles, it is possible to predict which cards are where and what is likely to come up next. Obviously, an enormous amount of pre-planning is required.

Here is where the computer comes in. "At first I would deal out up to 100,000 hands of blackjack on my kitchen table, analyse each and develop a strategy," says O'Brien. "This took me five years. However, last year I got hold of a program called the Universal Blackjack Engine. This can deal out all 100,000 hands in just under a minute. Using this, I was able to prove and disprove certain theories.

"Various casinos have different playing conditions - one deck, two decks and so on. So you can say: let's see what happens on a two-deck game, with three players at the table, and using this particular method of counting. You program all these variables in, then just press a button, and you can see what happens over millions of hands. From this, it's possible to determine whether a particular type of game is worth playing at a specific casino.

"Now I always travel with a laptop. Before I play each casino, in the hotel beforehand I check on the playing conditions and program them in. I'm therefore well prepared for every eventuality and can beat the system. In my opinion, every good card counter should be using a computer in this way."

At the end of the day, however, no one really gets rich except the bookies and the casino owners. Nevertheless, if a computer can help relieve them of just a small percentage of their profits, then it must be a good thing.