Bill Benter: The Mathematician Who Beat Hong Kong Horse Racing

This page contains affiliate links to ProfitDuel. If you subscribe through them we may earn a commission. How we make money.

Bill Benter is often described as the most successful gambler who ever lived. He never had a hot streak or a lucky tip. He built a statistical model, bet only when it said the price was wrong, and did it for decades in one of the deepest betting markets in the world.

From physics to blackjack

Benter was born in Pittsburgh in 1957 and studied physics. In his early twenties he read Beat the Dealer, Edward Thorp’s book showing that blackjack could be beaten by tracking the cards. Benter moved to Las Vegas and became a professional card counter.

Card counting worked, which was the problem. Casinos identified and banned successful counters, and Benter was pushed out of the game. He went looking for a market where a mathematical edge could be applied at scale without getting thrown out.

Hong Kong and Alan Woods

He found a partner in Alan Woods, an Australian professional gambler with a background in actuarial studies and blackjack. The two believed horse racing in Hong Kong had the right ingredients:

• Huge betting pools. Hong Kong racing attracted enormous turnover, so big bets wouldn’t move the odds too far.
• Pari-mutuel betting. Bettors bet against each other through a pool run by the Hong Kong Jockey Club, rather than against a bookmaker who could refuse their action.
• A closed system. A limited group of horses racing repeatedly on a small number of tracks produced consistent, well-recorded data.

They moved to Hong Kong in 1984. The early years were hard. The model had to be built, data collected by hand, and losses absorbed. After a profitable 1987 season, the partnership broke up, and the two went on to run separate operations.

How the model worked

Benter’s approach was to estimate each horse’s probability of winning, compare it with the odds in the pool, and bet on horses whose odds were longer than their true chance deserved.

The model used a regression technique suited to races, where exactly one runner wins. It began with a relatively small set of factors, including recent form, speed, jockey, distance, and draw, and grew over the years to more than 120 variables per horse.

The crucial step: blending with the market

Benter’s biggest insight was about the crowd. The public odds reflect things a model can’t see: stable gossip, how a horse looked in the paddock, inside information. Rather than ignore them, Benter combined his model’s probability with the probability implied by the public odds. The blended estimate was more accurate than either alone.

He later described this approach in a widely cited 1994 paper on computer-based handicapping. The idea is the same one modern +EV bettors use when they treat a sharp sportsbook’s line as the starting estimate and only bet when they have good reason to disagree. Our Pinnacle guide covers the modern version.

Kelly sizing

Benter sized his bets with the Kelly criterion, staking more when the model saw a larger edge and less when the edge was thin. Combined with the size of Hong Kong’s pools, this let a modest percentage edge compound into an extraordinary fortune.

The results

Benter’s operation became a data-and-computing business, running what amounted to a quantitative trading desk for horse racing. Bloomberg Businessweek reported in 2018 that he had earned close to $1 billion. He has also been a prominent philanthropist and, in later years, has been based in Pittsburgh.

His success helped legitimize computer-based betting and inspired a generation of quantitative syndicates in racing and sports.

What you can borrow from Bill Benter

You don’t need to build a 120-variable model to use Benter’s principles:

1. Respect the market. The public or sharp price already contains a lot of information. Start from it and look for specific reasons it’s wrong. That’s the logic behind devigging a sharp line.
2. Bet only on overlays. A horse is worth backing only when its price is longer than its true chance. The same is true of any sports bet. See positive EV betting.
3. Size by edge. Use a fraction of Kelly so bigger edges get bigger stakes and small edges get small ones.
4. Pick markets that won’t kick you out. Benter left blackjack for pari-mutuel racing, where winners weren’t barred. Today’s equivalent is using sharp books as your benchmark and managing sportsbook limits carefully.
5. Expect a long build. Benter’s early seasons lost money while the model improved. Any edge takes time and volume to show up.