Data Economy

The new Vegas bookie is a cyborg

Getty Images

There's an old saying among writers that there are only three human narratives: man versus man, man versus nature, and man versus himself. Add a fourth narrative for the age of big data: man versus the machines that are meddling in our daily lives.

In Las Vegas, where card counters from MIT and "rain men" have used peculiar powers of the human brain to cash out with millions, the man versus machine archetypal narrative is evolving as a plot line in the world of sports gambling. As another college football and NFL season gets underway, can the Vegas bookie who is like a stock character out of a Martin Scorsese movie still be the genius behind the three-and-a-half-point spread, or will he ultimately cede his place to a combination of Wall Street and Silicon Valley algorithms?

The somewhat surprising answer, according to sports analytics experts and statisticians turned sports-betting junkies, is that man will prevail, or at least give computer cores a run for their money. Even as algorithms moving the sportsbook line become more complex, human ingenuity and gut instinct can't be removed from the three-and-a-half-point equation.

So many variables can be plugged into algorithms now: player availability monitored minute by minute via social media posts, optical tracking of players on the field, and in-game predictive models projecting final scores based on situational scoring and defenses. It would be foolish to think casinos and sportsbook operations are not using every byte at their disposal to keep the edge in the house.

Cantor Gaming, the Vegas sportsbook affiliate of brokerage firm Cantor Fitzgerald, which brought Wall Street whiz to Vegas, declined to comment, but Vegas experts say Cantor has pushed the envelope the furthest in applying the algorithmic lessons of trading to sports betting.

(Read more: Nascar: No bottom line bang from big data)

Joe Peta, a 15-year Wall Street veteran who recently took a two-year detour into devising a baseball gambling system while recovering from a serious injury—an experience chronicled in his book, "Trading Bases"—said that he is convinced there will be many more intersections of analytics, investing and sports betting.

In baseball, the matchup between pitcher and batter reigns, and that dynamic makes it the most statistically friendly major sport on which people bet.

Nascar's new marketing strategy

There's a simple reason Las Vegas can't afford not to make the most of computing power: Sophisticated players such as Peta are looking for an advantage, and even short of true big data, algorithms based on core statistics can give it to them.

That was the experience of "Dr. Bob," Bob Stoll, who began applying his statistics background from the University of California, Berkeley to sports gambling in the mid-1980s, before the onslaught of big data number crunching.

"I had a huge edge doing what I consider to be now pretty simply analysis," Stoll said via email.

Otis Jennings, a statistics professor at Duke University's business school and an expert in sports analytics, said, "People out there are looking for an edge. ... The casinos need to do the same, and that's where sports analysts will get hired in Vegas."

(Read more: The Kit-Kat surveillance state)

Vegas is likely already well into that hiring spree, because the edge hasn't lasted, at least not to the degree it once did. In fact, Dr. Bob finds himself using more of his human brainpower. As really smart people doing complex math jumped into the market, the lines have followed, meaning that the books are more concerned with where the syndicates with computer models are likely to be betting than shading their lines toward public perception.

"I've actually had to change my approach a bit to take this into account, and my success now lies in being able to deviate from the numbers my models are cranking out and finding value in things the 'machines' aren't good at factoring in," Stoll said.

His percentage has come down—gone are the days of his legendary 60 percent-plus winning rate—but Dr. Bob has had success against the Vegas line in recent years. "I think there's still a place for the art of handicapping that the algorithms can't capture," he said.

Ed Feng, who runs The PowerRank website, which uses analytics to determine college football outcomes, appeared on the cover of Sports Illustrated last year predicting Alabama's victory over Notre Dame in the college football championship.

What he and other sport analytics experts do is beneath the level of big data, Feng said. A game like college football can be broken up into offensive and defensive matchups—passing versus a strong pass rush, a powerful running game against a vulnerable defensive line—and those kinds of factors can make teams that are even on paper look vastly different when ranked using a statistical model.

But that's still "little data," he said. "There is no need for petabytes of storage."

Feng, a Stanford Ph.D. who was on the academic track, decided to apply his acumen to sports when he saw what Google was doing with Page Rank and realized the algorithmic approach was a natural for the sports line. But he still believes "street smarts" have great value, especially in football.

"I think the race in sports gambling is less big data and more better algorithms," he said. "But it will never be at point where it's all computers—there is too much going on.

"You're not going to find a really true winning edge just looking at any ranking or trend, though you're probably behind if you're not using advanced computerized algorithms, and Cantor is pushing it," Feng said.

From the gambler against the house side, Feng cited Dr. Bob. "He is still winning at 57 percent. ... Bob does it the right way: he starts with the math and sees where it is going, but his intuition is the final decision."

Steve Millard, CEO of Kognitio, a data analytics company that works with international online sports gambling company bet365, said the casino industry is more likely to use big data for fraud detection than for sportsbook. But as social media becomes a major source of information (and not under the tight control of an organization such as the NFL) there may be more opportunities to gain from data analysis.

"Oddsmaking is an old-school craft," Millard said. "You get some sports junkies together and read newspapers and understand variables in the game, prior tendencies, emotions. … Football is a gladiator sport, and oddsmakers look at it with a human eye and human intuition. Machines can't make these decisions. Is there a better way to do it? Yes. Are we there yet? No."

Feng described the situation in terms appropriate for gambling. "If I were a smart sportsbook," he said, "I'd be interested, but I bet top gamblers are still ahead of them."

—By Eric Rosenbaum,