FEATURED MEDIA
Asset Pricing and Sports Betting
A bookmaker’s gambling ledger containing bets from thousands of sporting events seems more likely to inspire a plot point in a crime film than advance our understanding of the dynamics of financial markets. For Tobias Moskowitz, sports betting contracts were the ideal place to observe how prices shift due to overreaction to information.
In a novel approach, Moskowitz has used betting contracts as a new window to answer an old argument. It’s long been shown that a combination of value and momentum stock-picking strategies yields stronger-than-average returns. However, economists disagree over whether these high returns reflect some sort of risk premium, or a if there is a behavioral explanation. The Fama Family Professor of Finance at Chicago Booth realized betting contracts offered a way to test both cases. He shared his insights from the study at a Becker Brown Bag talk to MBA students on February 24, 2014.
Moskowitz realized that gambling data has unique features that make linkages between markets, aggregated risk, and models with limited scope less of an issue than with standard financial data. “What’s nice is that betting contracts are independent and purely idiosyncratic, not only to the market, but also to each other,” said Moskowitz. “Certainly between sports and even within a single sport, it’s very rare that one game would influence another, apart from rare situations.”
He analyzed 120,000 gambles across 18,681 NBA games, 7,035 NFL games, 23,986 MLB games, and 9,890 NHL games, examining three points in the lifecycle of each bet: the opening price set by the bookmaker, the closing price before game time, set by betting activity, and the price at the close of the game, when the payoff is determined by the final score. Examining the regression of the betting horizon between each price, Moskowitz hypothesized that the price would move because for two reasons: shifting sentiment and new information. But what he didn’t know for certain was which of the two would be a better predictor of the price of a gambling contract.
Suppose that betting markets are governed solely by the efficient market hypothesis, Moskowitz noted. “A key player gets injured. You thought that the spread ought to be three points, but suddenly Peyton Manning isn’t playing, and the spread drops.” Under an efficient market, the effect of that new information on the price of the bet would be near instantaneous, and very little change in price would happen between the game starting and ending.
However, under a more behavioral model, sentiment may take over. “People get excited about a team, they like the color of the jersey, whatever it might be, but there’s no information content. Then when prices move [during a game], what’s going to happen?” asked Moskowitz. “If prices move for non-information reasons, they have to be corrected by the time the game ends, since the game outcome is not affected at all by betting behavior.” That means if gamblers feel strongly for a team before a game and overprice bets on their victory, the price will fall as the game’s outcome brings that price more in line with the actual payout of the bet; if they underestimate and undervalue a team’s chances for success, the game outcome will raise the price of a bet on their victory.
So which view pans out in the data? “When the betting opens, people chase returns, and they push up the price, very much like what the behavioral theory suggests happens in financial markets,” says Moskowitz. “Perhaps more interestingly, it reverses by the time the game ends; the game outcome is unrelated to the information content related to chasing returns. What that means is that betting contracts are overpriced [at the start of the game], and you’d be better off placing bets when betting opens.”
The statistical significance of this trend holds more strongly over the short term than the long, meaning the behavioral explanation for price fluctuations isn’t as compelling if you were planning on pursuing a value investment strategy compared to chasing momentum-based returns on your bet. “It’s a question whether you think the glass is half-full or half-empty. The patterns look somewhat compelling, but the statistical significance is slight,” says Moskowitz.
In this light, Moskowitz says the work affirms many behavioral results found in financial markets. But, he cautioned hopeful MBAs in the audience, the potential of leveraging this research is limited. The 10 percent “vig,” or transaction cost for a bet, cuts deeply into the profits statistically possible using the study as a guide, limiting the study’s potential for a usable strategy to pay tuition bills with gambling winnings.