Can you ever imagine yourself winning a billion dollars? Okay, not a billion Australian dollars but a billion Hong Kong dollars? This is about $200 million in Australian currency. Sounds like a fantastic pipe dream, doesn't it?

Let me assure you it isn't. Some Aussies, Canadians and Americans are winning massive amounts of money betting on racing in Hong Kong, and one group says it has passed the billion-dollar mark in profits!

They bet via carefully programmed computers and a bet on one race will probably average around $320,000. They win most of the time. Their ingenious computer rating system has made them super-rich.

Obviously, in a story like this there can be no names. They want their anonymity. They have an office, but it's a shabby one which you'll find somewhere in the teeming streets of Kowloon, well away from the tourist nightspots. On racedays in Hong Kong, the super bettors swing into action.

The billion-dollar partnership, run by two Americans, is now said to be the most successful to bet on racehorses since the racing game began.

The leader of the betting blitzers calls himself Mike. He studied physics at Bristol University in England, then read a book called Beat The Dealer, by Ed Thorpe, and realised that Thorpe's maths had the potential to help a bettor win a fortune.

Thorpe's book was all about blackjack, and Mike tried his luck, with a bank of just $200, at Atlantic City, New Jersey in the late '70s. He lost the $200 but then had a chat with his parents and friends, and took a year off his studies to pursue his dream in Las Vegas.

He had a bank of $500, obtained from working in a fast-food shop for six months. In his spare time he practised and tried to refine Thorpe's thinking. In Las Vegas, he was joined by two friends and they pooled their combined savings, a total of $2000.

The problem with the casino approach, says Mike, was that the stake had to be increased by a factor of up to 10 when high cards were anticipated and that made their identity "pretty transparent" to the casino bosses, who were always on the lookout for card-counters.

Twelve months after trying his hand in Las Vegas, Mike gave up gambling on the Thorpe method but decided to continue his studies using computers, trying to write a program that would identify the bias in a roulette wheel. It was at this time that an Australian blackjack team hit town.

Short of a couple of players, they asked Mike to join them. But the casinos were virtually on a war-footing against the card-counters and imposed shoes of six decks, which were dealt until only a third of the cards had been used, and then the shoe was shuffled again. The card counters were stymied.

But Mike's eyes were sharp and he soon began to observe the various shuffles at the different casinos. He then wrote a computer program that registered the fall of the high and low cards, and the type of shuffle used, but also signalled when a run of high cards could be anticipated.

Amazingly, the player sat at the blackjack table with a small computer strapped to a leg, connected by a wire through the groin area to a battery strapped to the other! Each big toe had switches which connected above and below, and these were used to input data. When conditions were favourable the computer produced vibrations which instructed the player when and how much he should bet.

But, like all good things in casinos, the ruse was tumbled - but not before Mike and his team had netted some $US8 million. The casinos became so upset about electronic devices that Nevada passed a law to make them illegal.

Mike began to look abroad for betting opportunities. The story goes that an Aussie named Peter knew horse-racing was booming in Hong Kong, so in 1984 Mike and Peter headed to Asia to try their luck.

Firstly, they knew they had to have a form database. They employed girls to type four years of form information into the computer and then began the task of trying to make sense of Hong Kong racing form.

The theory they used was a simple one: Devise a method that would accurately estimate the chance of each horse in a race. These odds would then be measured against the odds offered on the Hong Kong tote, the team betting when an overlay was available.

Mike read work by two academics, Ruth Bolton and Randall Chapman, entitled Searching For Positive Returns At The Track, a Multinomial Logic Model For Handicapping Horse Races. By using a version of their 'multiple regression analysis', Mike then tried to identify and measure the various factors that affected a horse's performance. From there he would calculate the correct weighting for each factor.

Nothing too surprising in all that. Most good bettors operate in a similar manner. The results from their first few years of operation were modest, and in 1987 the team split, with the Aussies who had joined going their own way.

Mike then teamed up with a Canadian named David, who was a brilliant computer programmer. They devised a far more sophisticated method. They even enlisted the help of an American professor of mathematics, and learned about the relationships between performances and other race factors using a 'curve' method of analysis. Their computer today examines 120 factors to determine a horse's performance merit.

Once they learned how to 'rate' the horses well ' they had to learn how to bet them to best effect. They now use the Kelly optimal betting approach.

Says Mike: "We found that if we gave a horse, say, a 20 per cent chance of winning, and the public odds reflected a 10 per cent chance, then the true odds were nearer to 16 per cent."

The team readily concedes that note has to be taken of the 'inside information' that is reflected in the betting pool (which is huge in Hong Kong). Also, a computer analysis of jockeys was found to be of utmost importance.

"We have found it to be one of the most important variables," explains Mike. "And it could be broken down into two separate constituents. We estimate the physical ability of a top jockey to 60 per cent; the other 40 is what might be regarded as a signal from the stable of their intentions when they booked him."

Essentially, the team's computer calculates a horse's odds, reviews the prices created by the betting public, adjusts accordingly, and then scans all the different forms of betting available, searching for value betting opportunities.

When it identifies bets that are available at odds significantly higher than its own, it then calculates the size of the bet to be placed. This final part of the approach is vitally important.

The greater the differential, the larger the bet. But the bet has to be of a size that does not affect the dividend too much, otherwise the team's edge disappears!

Mike says the team wins on two meetings out of three, and they are always trying to streamline the selection approach. If they want to test a refinement it can be done immediately.

The computer tells them in a matter of seconds if the modification is better than what they already have.

Now for the figures from their betting: Because of the huge pools in Hong Kong the big punters there are several others apart from Mike's group - can lay out enormous amounts and not make a dent in the pool.

The record turnover for a single race in Hong Kong is around $40 million. But even Mike is looking elsewhere. Recently he was in Japan and describes its betting pools as 'huge' with the Japan Cup attracting turnover of $900 million.

Mike and his partner, David, now have more money than they will ever be able to spend. They've even established a charity foundation so they can share some of the spoils.

They've been in business for some eight years now. They have turned over around $HK2 million a race. Their profit, they admit, is past the billion-dollar mark.

Their closest rivals, they say, are Australians, a secretive group who bet big, though not as big as Mike and David.

Mike's one criticism of the Hong Kong betting set-up is that sending bets into the Jockey Club's system is relatively slow at six per minute, although multiple combinations do count as one bet and may involve several hundred individual wagers.

The final moments before a race starts are hectic and full of stress.

Says Mike: "You get a lot of horses that are over-priced in the win pool and assume that if they are involved in the finish you will win but the pools occasionally have imbalances and there are times when we have been surprised not to have won.

"But then again we've been surprised the other way, too."

He says he is amazed at the tenacity of ordinary punters who try to analyse races without the help of a computer.

"Starting from scratch it's an heroic effort every day," he says. "The great beauty of a computer is that all the knowledge is stored and available at any moment for you."

It's certainly working for him. He has a billion Hong Kong dollars to prove it.

By Brian Blackwell