The Octopus Is Dead, and Big Data Is the Future of Sports Gambling
by Adam Tanner
In 2010, a German octopus named Paul received worldwide attention for correctly forecasting the final matches of the World Cup. But this year, following Paul’s death, data scientists at companies such as Google and Microsoft gained the spotlight by accurately predicting the winners using big data.
Google correctly called 14 out of 16 of the final matches. One of the calls missed Germany’s 1–0 victory over France in the quarterfinals. Microsoft crunched numbers from past games and player performances to predict the winning team in 15 of 16 final matches (they erred in thinking Brazil would beat the Netherlands in the consolation game for third place).
Of course, fans have long closely followed sports statistics, but big data has created a new dimension of detail that can provide unprecedented insights about what may happen in the future. “data…always existed, but now it is supercharged,” says David Rothschild, an economist who developed the predictive model that Microsoft used for the 2014 World Cup. “We have the ability to look at data flows, rather than looking at them in the box score the day after in the newspaper. You are actually updating with extremely low latency in real time.”
In recent years, a growing number of math experts have entered the field of sports betting, convinced that data on past results, location of game, weather and various player statistics can give them a unique edge.
Once upon a time, someone like Paul Bessire might have gotten a job on Wall Street. Instead he spends his time gambling on sports and sharing his insights on PredictionMachine.com.
“In the financial world I am competing against everybody else, thousands of people. At that time Goldman Sachs or Lehman Brothers had all the same kind of models and information that I did and the data that I did, whereas in sports it was very different and remains very different,” he says.
A factor that is unique to sports betting is that many will relentlessly wager on their favorite team, rather than use dispassionate analysis. Bessire says that “Basically that is what will keep this an exploitable market.”
“I live in Cincinnati. My neighbor does not buy Procter & Gamble stock because we live in Cincinnati. He might buy it for other reasons but he is not going to go out and buy it because he thinks out of loyalty he should. He would bet on the Bengals, just because we are in Cincinnati and the Bengals are playing and he wants to root for his team.”
Yet even with the power of big data, veterans such as Bessire often forecast wrongly, as there are a lot of random variables that can change the outcome of a game. He says he correctly calls nearly 55 percent of NFL football games. That means he is wrong 45 percent of the time, but he is correct more than the 52.4 percent of the time needed to come out ahead in Las Vegas sports gambling given the house’s take.
Matt Holt, vice president of business development at CG Analytics, says the ever-expanding wealth of sports statistics is a double-edged sword for his firm, which sets the odds in major Las Vegas casinos including the Cosmopolitan and the Venetian, and other bookmakers.
“Information travels at a speed that it never has before,” he says. “As soon as any single player on the team even breaks a nail, there’s five people tweeting about it, everybody knows how any player is going to miss a game, his girlfriend just got pregnant or the weather is going to be a tornado or he might get suspended for smoking weed.”
Holt and other Vegas insiders I spoke with were less impressed with the Google and Microsoft’s successes. “The favorites were coming in left and right. It was just the better team beating the weaker team on paper,” says Raphael Esparza, a veteran handicapper at Doc’s Sports Service.
In fact, he says for all the talk about big data, many Vegas bookmakers have not changed their traditional approach. “Everyone thinks it’s all mathematical and we have a guy that’s in a locked room that does statistics and all that,” Esparza says. “There is really no number crunching, no analytics. It’s just knowing your craft and knowing the sport is probably, I would say, 80 percent of how you make the numbers.”
“I was in the industry for 14 years and there’s maybe a handful of days that we got our butt handed to us. And it’s not because of statistics or because somebody found a way, or an octopus told it to bet it. It was just that we lost that day.”
David Rothschild of Microsoft Research concedes that rather simple information can very often predict the future in sports. “What is the most surprising thing is how accurately you can forecast sports with two statistics in every sport: the points for and against and home and away,” he says. “It doesn’t matter what sport I am talking about as long as the sport is won or lost on some basis of points.”
But Theodore Todorow, a former algorithmic developer at Cantor Gaming (now CG Technology), says the more instinctual odds makers are a dying breed. “Yes, there is this old time mentality but it is a bit like on Wall Street,” he says. “The person with the data has the edge.”
For more by Adam Tanner, check out:
What Stays in Vegas: The World of Personal Data—Lifeblood of Big Business—and the End of Privacy as We Know It, published September 2, 2014 by PublicAffairs.