The Perfect Bet

by Adam Kucharski

  With soccer data hard to come by in the late 1990s, gamblers had to get ahold of the information in any way they could. In some cases, people created automated programs that would scour the handful of websites that did publish results and copy the data tables straight off the webpages. Although this “screen scraping” provided a source of data, the websites being scraped did not like gamblers taking their content and clogging up their servers. Some installed countermeasures—such as blocking certain IP addresses—to stop people taking their data.

  Even in the data-rich world of US sports, there is still plenty of variation in information between different leagues. One of the reasons Kent analyzed college sports was the amount of information available. “There are so many more games in college basketball, and a lot more teams,” Kent said. “You get a huge database.” Having access to these data helped Kent predict the outcomes of matches and place appropriate bets beforehand.

  THROUGHOUT KENT’S CAREER, SPORTS betting in Las Vegas would come to a halt once a game started. By the time the referee blew the whistle to start the match, Kent’s money was already down. The gambling and the action, two things that seemed to be so closely linked, were instead separated. It wasn’t until 2009, when a new company arrived in the city, that casinos finally fixed this broken Venn diagram of gambling. That company was Cantor Gaming, part of the Wall Street firm Cantor Fitzgerald. In recent years, it has become the resident bookmaker at a number of major casinos. Walk into the sports section of the Venetian, the Cosmopolitan, or the Hard Rock, and you’ll find dozens of big screens and betting machines, all operated by Cantor. Crammed between coverage of everything from baseball to football, there are rows of numbers and names, showing the odds for different matches. These “betting lines” rise and fall with the noise of the crowds. The room feels like a hybrid of a sports bar and a trading floor, a place where drinks and data blend together under the eternally bright casino lights.

  The numbers on Cantor’s screens might reflect the emotions of the spectators, but they are actually controlled by a computer program that adjusts the betting lines throughout the course of the game. Cantor calls it the “Midas algorithm.” If something happens in the game, the program updates the odds on display automatically. Thanks to Midas, in-play betting has taken off in a big way in Las Vegas.

  Much of the credit for the Midas software goes to an Englishman named Andrew Garrood, who joined Cantor in 2000. Before that, he’d been working as a trader for a Japanese investment bank. The leap to Las Vegas wasn’t as big as it might seem: Garrood simply went from designing models that could price financial derivatives to ones that could put a value on sports results.

  Cantor’s biggest statement of intent came in 2008, when it bought a company called Las Vegas Sports Consultants. This company came up with odds for bookmakers across Nevada, including nearly half the casinos in Las Vegas. Cantor wasn’t just interested in its predictions, however. In buying the company, Cantor had secured an extensive database of past results for a whole range of sports. The information would form a vital part of Cantor’s analysis. From baseball to football, Cantor needed to know how certain events changed a game. If the San Francisco Giants get another home run, how would it affect their chances of winning? If the New England Patriots have one last attempt to get the ball down in the dying moments of a game, how likely are they to pull it off?

  According to Garrood, straightforward “vanilla” events are relatively easy to predict. For instance, it’s not too difficult to work out the chances a football team will score a touchdown if they start a drive from the 20-yard line. The problem is that there could be many successes and failures during a game, some of which are subtler than others. Which events are worth knowing about? Garrood has found that most plays don’t affect the outcome much. It’s therefore important to pin down the crucial events, the ones that do make a big difference. This is where the enormous database comes in handy. While many gamblers are relying on gut instinct, Midas can assess just how much effect that touchdown will really have.

  How does Cantor make sure it gets all its predictions right? The answer is that it doesn’t try to. There is a commonly held view that firms like Cantor use models to try to nail the correct result for every game. Matthew Holt, director of sports data at Cantor, has rebuffed this myth. “We don’t make lines to predict the outcome of a game,” he said in 2013. “We make lines in anticipation of where the action will come.”

  When it comes to betting, bookmakers’ aims are fundamentally different from those of gamblers. Suppose two tennis players in the US Open are perfectly matched. The game is 50/50, which means that for a $1.00 bet, a fair return would be $1.00: if a gambler bet on both players, the bettor would come out even. But a bookmaker won’t offer odds that return $1.00. Instead, it might offer a payoff of $0.95. Anyone who bets on both players will therefore end up $0.05 poorer.

  If the same total amount is wagered on each player, the bookmaker will lock in a profit. But what if most bets go on one of the players? The bookmaker will need to adjust the odds to make sure it stands to gain the same amount regardless of who wins. The new odds might suggest one player is less likely to come out on top. Smart gamblers, who know that both players are equally good, will therefore bet on the one with longer odds. For bookmakers that have done their job properly, this isn’t a concern. They don’t move their betting lines to match the true chance of a result happening. They do it to balance the books.

  Every day, the Midas algorithm combines computer predictions with real betting activity, tweaking the odds as wagers come in. It performs this juggling act for dozens of different sports, simultaneously updating betting lines as games progress. To make a profit, bookmakers like Cantor have to understand where gamblers’ money is going. What are they betting on? How might they react to a particular event?

  Just as information flows between bettors and bookmakers, in many instances gamblers will also try to work out what their rivals are doing. When word gets out that a betting syndicate has come up with a successful strategy, others inevitably want a piece of the action. Because many betting strategies have their origins in academia, it’s often possible to piece together the basic models by sifting through research articles. But sports betting is a competitive industry, which means some of the most effective techniques remain shrouded in secrecy. According to sports statistician Ian McHale, “The proprietary nature of prediction models means that the published ones rarely (if ever) represent the very best models.”

  If gamblers don’t know who has the best strategy, it can create a tense environment. In the huge Asian markets, where many of the largest soccer bets are placed, wagers are often made via instant messaging software. At the same time, information ricochets between bookmakers and gamblers, as each tries to work out what the other is thinking and how the other will bet. “The betting grapevine is huge,” as one industry insider put it. “There is a lot of paranoia.”

  WHEN ASIAN BOOKMAKERS GET coverage in the Western media, it’s not usually for a good reason. After some suspicious bowling in cricket matches between Pakistan and England in 2010, three of Pakistan’s players were handed bans for agreeing to deliver bad balls. Reporters noted that bookmakers—many of whom were based in Asia—would often target such games. The scandals have since continued. During the summer of 2013, three cricketers playing in the Indian Premier League were charged with match fixing. Police claimed bookmakers had promised upward of $40,000 to the players if they let the opposition score runs at specific times. Then, in December 2013, UK police arrested six soccer players for allegedly offering to collect yellow or red cards to order.

  There is certainly a huge appetite for gambling in Asia, and not all of it aboveboard. It’s been estimated that the illegal betting market in China is ten times larger than the legitimate sums handled by the Hong Kong Jockey Club. Illegal betting is also common in India. When the national cricket team plays arch rival Pakistan, the total amount wagered can approach $3 billion. Yet the
Asian betting market is changing. Gamblers no longer need to track down black market bookmakers in rooms behind backstreet bars. There was a time when they would need to bring cash and a code word; now they can bet by phone or online. Glossy call centers have replaced grimy betting rooms. The new industry is a step away from the illegal black market, but it remains little regulated. This is the “gray market”: modern, corporate, and opaque.

  When it comes to high-stakes betting on sports such as soccer, Asia is the location of choice for many Western gamblers. The reason is simple. In Europe and the United States, bookmakers rarely take large bets. As a result, gamblers based in these regions are finding it harder to stake the sort of money that will make their strategies profitable. Despite being a prolific bettor—or rather because he is a prolific bettor—Haralabos Voulgaris has complained that US bookmakers are reluctant to take his bets. Even when they do, the betting limits are placed at unhelpfully low levels; he might be allowed to stake only a few thousand dollars. Not all Western bookmakers shun successful bettors, however. In the past decade, one firm has gained a reputation for accepting—and even encouraging—bets from smart gamblers.

  When Pinnacle Sports started in 1998, it was clear that it had some bold ambitions. The betting limits were high, with maximum stakes larger than those offered by many existing bookmakers. Pinnacle claimed it was happy to let players bet the maximum as often as they liked. Even if a player consistently made money, Pinnacle wouldn’t shut the player down. Back in 2003, such ideas went completely against established bookmaking wisdom. If you want to make money, went the dogma, don’t let smart bettors place huge bets. And certainly don’t let them do it again and again. So, how did Pinnacle pull it off?

  Whereas all bookmakers look at overall betting activity, Pinnacle also puts a lot of effort into understanding who is placing those bets. By accepting wagers from sharp bettors, Pinnacle can get an idea of what these gamblers think might happen. It’s not very different from Bill Benter combining his predictions with the public odds displayed at Happy Valley. Sometimes the public knows things that a betting syndicate—or a bookmaker—might not.

  Pinnacle generally posts an initial set of odds on Sunday night. It knows these numbers might not be perfect, so only take a small amount of bets at first. It has found that the first bets almost always come from talented small-stakes bettors: because the early odds are often incorrect, sharp gamblers pile in and exploit them. But Pinnacle is happy to hand an advantage to these so-called hundred-dollar geniuses if it means ending up with much better predictions about the games. In essence, Pinnacle pays smart gamblers for information.

  The strategy of purchasing information has been attempted in other walks of life, too, sometimes with controversial results. In the summer of 2003, US Senators stumbled across a Department of Defense proposal for a “policy analysis market” that would allow traders to speculate on events in the Middle East. It would be possible to bet on events such as a biochemical attack, for example, or a coup d’état, or the assassination of an Arab leader. The idea was that if anyone had inside information and tried to exploit it, the Pentagon would be able to spot the change in market activity. Investors might make a profit, but they would also reveal their hand in the process. Robin Hanson, the economist behind the proposal, pointed out that intelligence agencies by definition pay people to report unsavory details. In moral terms, he didn’t see a market as any better or worse than other types of transactions.

  The Senators disagreed. One called the idea “grotesque”; another said it was “unbelievably stupid.” According to Hillary Clinton, the policy would create “a market in death and destruction.” The proposal did not survive long in the face of such fierce opposition. By the end of July, the Pentagon had scrapped the idea. The decision was arguably an ethical rather than economic one. Although critics attacked the morality of the proposal, few disputed that betting markets can reveal valuable insights about an event. Unlike participants in an opinion poll, gamblers have a financial incentive to be right. When they make predictions about the future, they are putting their money where their mouth (or model) is.

  Today, Pinnacle canvasses gamblers’ opinions on a wide range of subjects. People can bet on the identity of the next president or who will take home an Academy Award. Pinnacle has so much faith in the approach that it regularly takes large bets on popular events: in the past, it has been possible to wager half a million dollars on the soccer Champions League final. Because Pinnacle’s business model relies on having accurate predictions, there are some things it doesn’t take bets on. In 2008, for instance, Pinnacle dropped horse racing as a betting option because it doesn’t specialize in the sport.

  Companies like Pinnacle, which have found a way to combine in-house statistical predictions with the opinions of smart gamblers, have challenged traditional bookmaking. By harnessing the knowledge of smart gamblers, they have more confidence in their odds, and hence are happy to take larger bets. Yet bookmakers are not the only ones changing. In some cases, gamblers are skipping the bookmaker altogether.

  DURING THE PAST DECADE or so, approaches to betting have changed dramatically. As well as wagers moving online, bookmakers have faced competition from a new type of gambling market, in the form of the betting exchange. This is much like a stock exchange, except instead of buying and selling shares, gamblers can offer and accept wagers. Perhaps the best-known betting exchange is the London-based Betfair, which handles over seven million bets a day.

  Betfair’s creator, Andrew Black, came up with the idea for the website during the late 1990s, when he was a programmer at the British Government Communication Headquarters in Gloucestershire. Security wouldn’t let him stay on site past five o’clock, so he found himself spending each evening alone in his rural farmhouse. Having so much free time was a burden, but it was also rather fruitful. “The boredom was horrendous,” he later told the Guardian, “but mentally I became really quite productive.”

  While attending college, Black had developed an interest in betting. But there were some drawbacks with the traditional way of gambling, and during those evenings in Gloucestershire, Black thought about how things could be improved. Rather than going through a bookmaker, as he’d always had to do, why not let gamblers bet directly against each other? The project meant combining ideas from financial markets, gambling, and online retail. Black, who had previously spent time as a professional gambler, stock trader, and website builder, had experience in all three of these areas.

  The Betfair website launched in 2000. That summer, the company arranged for a mock funeral procession to pass through the city of London, with a coffin announcing the “death of the bookmaker.” Although the stunt brought plenty of media coverage, competitors were already lurking. One rival website mimicked eBay: if someone wanted to place a bet of £1,000 at certain odds, the site would try to pair the bettor up with someone happy to take such a bet. Trying to pair people up was a bit like playing a giant online game of snap. And that sometimes meant waiting a long time for a match.

  Fortunately, Betfair had a way to speed things up. If there were no takers for a wager, the website would divide up the bet between several different people. Rather than trying to find someone willing to take the full £1,000, for example, the website might slice up the total and match it with, say, five people wanting to accept a £200 wager. Whereas bookmakers had traditionally made their money by tweaking the odds on display, Betfair left the odds untouched and instead took a cut from the profits of whoever won a particular bet.

  BETTING EXCHANGES LIKE BETFAIR have opened up a new approach to gambling. Unlike traditional bookmakers, you aren’t limited to betting for a particular result. You can also “lay” the result by taking the other side of the bet; if the result doesn’t happen, you win whatever was staked.

  Because you can bet both ways on a betting exchange, it’s possible to make money before a match ends. Suppose a betting exchange currently displays odds of 5 for a particular team. You decide
to bet £10 on them, which means you’ll get back £50 if they win. Then something changes. Perhaps the opposition’s star player picks up an injury. Your team is more likely to win now, so the odds drop to 2. Rather than waiting until the match ends—and risking the result going against you—you can hedge your original bet by accepting someone else’s £10 bet at the lower odds. If your team wins, you’ll get £50 from the first bet but have to cough up £20 for the second; if your team loses, the two bets will cancel each other out, as shown in Table 4.1. The match hasn’t even started and you’re guaranteed £30 if your team wins, and you won’t lose anything if they don’t. (Many bookmakers have since introduced a “cashout” feature, which in essence reproduces these trades.)

  Because you can back and lay each result, the Betfair website displays two columns for every match, showing the best available odds on each side of the bet. Such technology has made it easier for gamblers to see what others are thinking and to take advantage of odds they believe are incorrect. Yet it’s not just bookmaking that is becoming more accessible.

  TABLE 4.1. Bets Can Be Hedged by Backing and Then Laying the Same Team

  SCIENTIFIC BETTING STRATEGIES HAVE traditionally been the preserve of private betting syndicates like the Computer Group or, more recently, consultancy firms like Atass. This may not be the case for much longer. Just as banks offer clients access to investment funds, some companies are letting people invest in scientific gambling methods. As Bloomberg columnist Matthew Klein put it, “If I find a guy who is good at sports betting and is willing to bet with my money in exchange for a fee, he is, for all intents and purposes, a hedge fund manager.” Rather than putting money into established asset classes such as shares or commodities, investors now have the option of sports betting as an alternative asset class.