When Gareth Morgan was a young student living in a bus, he had an idea that he thought would make him a millionaire.
"My sister-in-law said: 'Oh Gareth, you spend so much time on computers and numbers. You'd think you could tell me who's going to win the second leg of the double.' And I thought, 's***, I can'."
Morgan put his PhD in economic modelling to work on all 7000 racehorses in the country, with an army of trackside informants measuring their race times. He raised $1 million - "which is amazing, when you think I had the bum hanging out of my jeans" - and launched a punter's guide, called Bettor Informed.
The theory was simple. At most big meetings in Auckland and Wellington, the TAB got the odds more or less right because the horses' form was well known. But at smaller country meetings the market was less efficient. All punters had to do was concentrate on these races and put their money on horses where the TAB offered a bigger payout than the guide's estimated fair dividend.
Morgan, who needed 14 per cent market share to make a profit, quickly learned most racing fans were more interested in the number of winning horses they backed than their overall profit. Yet to make his system work, you had to ignore low-priced favourites and refuse to bet at major meetings, "which was no fun at all".