字幕列表 影片播放 列印英文字幕 The Happy Valley and the Hong Kong Jockey Club, it's kind of a Mecca for horse racing. People there bet more in terms of their income than almost anywhere else on the planet. In 2001 a very large jackpot had built up. Roughly 20 million dollars U.S. One million people placed a wager on the Triple Trio for that race. That's equivalent to about one in seven of the population of Hong Kong at the time. Last leg of the Triple Trio and away they go The Triple Trio is three races, the top three finishers in any order in all three races, the odds are something like 10 million to one. Bobo Duck in front, Mascot fighting back, Bobo in front and there's the training double for Derek Cruz. No one came forward to claim the prize. The winner was an American professional gambler. At that point he was making so much money from his gambling activities that it didn't matter to him. This is the story of a shy computer nerd who came up with a system to beat horse racing and made a billion dollars. My name's Kit Chellel. I'm a reporter for Bloomberg and a writer for Businessweek. The fundamental unfairness of sports betting is that over a period of time, you will almost always lose if you make a lot of bets. Gambling, sports betting is slanted against you as a punter, as a gambler. Bill Benter decided to beat that system. He just became captivated with this idea that these games, which supposedly casinos had this long standing advantage, you could actually beat with the right methods. I'm Adam Kucharski. I'm a mathematician and science writer. When he was younger he saw a sign in Atlantic City that said "card counters are banned from our games" and for him that sign basically said card counting must work. He picked up a book by Ed Thorpe, called Beat The Dealer, which outlined the fundamentals of card counting. You're looking for how many high or low cards have emerged and how many might remain in the pack. If you think there's a high chance that there are lots of high cards left in the pack, that changes the odds in your favor. Lots of people went and tried card counting. Bill was more successful than most, managed to make some money and make a career out of it. He teamed up with Alan Woods, an Australian professional gambler, but as is always the problem with professional gamblers, casinos don't like people who win. And they were systematically excluded from all the major casinos around Vegas. So they needed to find a new game. Alan Woods was interested in horse racing and between them they talked about it and became attracted to the betting scene in Asia and in particular the Hong Kong Jockey Club where they were aware there was this huge horse racing market they might be able to tap into. One of the reasons that Hong Kong has a big advantage if you're trying to use mathematical and statistical methods is it's almost like a laboratory for studying horse racing. In Hong Kong, you've got two race courses that dominate. You've got a huge amount of betting activity. You've got a pool of about a thousand horses. You've got hundreds of races a year. So if you're designing something that's going to generate reasonably consistent data to study these patterns, Hong Kong is the place that you want to go. So they sent Alan Woods to Hong Kong and he came back with all the results of racing in Hong Kong going back five, ten years I think. And they manually fed that into a very basic early computer to try and create a statistical model. They fed the numbers into their computer. Took them a long time. It took them about nine months. So what they would do is use these measures of horses' quality and come up with a prediction of each horse's, in a given race, their chances of winning. And then they would compare that to what the odds on the board implied those horses of chance of winning and what you want is a horse where the odds imply it has a very low chance, but your model says it has a pretty decent chance. In the course of researching this story I spoke to gambling experts they said it was impossible, couldn't be done. If you bet long enough on horses, you will lose money. There are lots of factors that go into a horse winning or not. The favorite doesn't always win. In his first year of betting, I think he lost $150,000. It had all sorts of glitches because their data wasn't perfect. The fact that the Jockey Club takes a 15, 17% cut makes it harder to be sustainably profitable. It means you have to be not just better than the market, but you need to be about 20% better than the market to start making a profit. In a parimutuel system, in this case the Hong Kong Jockey Club, the odds are determined by the action. So the number of people betting on each particular horse, and the sums they bet, determine the odds. And what they did, alongside developing predictions, was find ways of really carefully trickling their money into the market if there's a certain amount in the pot and you go and put a huge amount on a particular horse, you're actually going to affect how much money you can make back because you're going to swing the market so much towards a particular horse. Eventually he came to incorporate the public odds into his model and that was a huge breakthrough for him. He'd worked out that the individual decisions of a million punters in Hong Kong were actually a fairly accurate predictor of a horse's chance of success. Bill Benter has been betting for a few years in Hong Kong and he's starting to make serious money, enough that he can hire employees to place bets over the phone for him. He buys a load of computers, he hires a nice office. It was really operating like a business and much like you'd see the sort of actions in a modern hedge fund in that they were accumulating really good data, they were analyzing that data and then they were developing predictions that they could act on. And really the nature of this business, these syndicates, it wasn't that they were placing the odd bet and then celebrating. It was just almost a systematic trading role where they were expecting to make money over time. One day he gets contacted by the Hong Kong Jockey Club and as a professional gambler who's worked in Vegas, I think he assumed it was going to be bad news, they might try and kick him off the betting pool. And instead they said "You're one of our best customers. What can we do to make your life better?" Because they take a percentage of the total betting pool, the more computer bettors there are, the more money they're betting, the more profit they make. Bill wanted a computer and they gave him one. They gave him something called the 'Customer Input Terminal' and allowed him to stop having to have employees pick up the phone and spend 10, 20 seconds each time placing a bet. He could do it in a fraction of a second. So it allowed him to make more bets. The more bets you make, the more likely it is that your edge pays off. So when he was suddenly able to make thousands upon thousands of bets at the click of a button, it dramatically increased his profitability. There grew this kind of subindustry of specialists. I don't think the Hong Kong betting public had broad awareness that there were computer experts using statistical modeling to bet in the same pools that they were. They simply weren't well known. They never appeared in the press. No one knew the name Bill Benter. The amount of money that was coming out of the pool and going to computer teams just grew and grew and grew and grew. The computer betting teams got too good. They were making too much money, they were making so much money. The Jockey Club saw what was happening and they realized that if word got out, that the specialist computer teams, foreign-owned and run, was siphoning money out of the betting pool to the detriment of the average bettor, it might discourage people from going to the race track and making bets. They abruptly cut off Bill Benter. But then this record jackpot came along. I think they made 50,000 bets, they spent well over a million Hong Kong dollars, they went into it deciding that even if they won, they weren't going to claim the money. They almost did it to test how good their model was by this point. Does Bill Benter fit into the stereotype of a professional gambler? No, hardly at all. He had more than 30 winning combinations, but he had one single winning ticket that correctly predicted all nine placed finishers. And that was the one that had the big payout. There was a purity about what he did, you know, he was faced with a technical, numerical challenge. His job was to beat the odds in horse racing and the way he did it was applying science to this problem in a way that most gambling people don't. He's still working away at that problem, you know, the algorithm is constantly evolving and changing and improving. Bill was really one of the first people to apply computing power and technology to a system of odds-making. But that practice is now spread all over the world, not just by bettors but by bookmakers too. So in a sense he transformed the industry. His obsession is the science of gambling and probability and to this day it's what he does.