Betting Against Efficiency: Behavioral Finance in an Nfl Gambling Exchange

Abstract

Although it may be a stretch to claim that the efficient markets hypothesis of Eugene Fama (1970) has been toppled, recent work has uncovered both empirical and theoretical opposition to Fama's pillar of financial theory. Much of this substantial body of research has focused on the world's financial markets. The explosion of internet-based betting exchanges and information markets, however, provides a rich and simple dataset with which to examine market behavior. These markets provide an especially interesting environment under which to study pricing behavior because the fundamental value is always revealed at the close of each contract, unlike stock prices that merely provide repeated estimations of fundamental value. Concentrating on trading in contracts with the outcomes of National Football League (NFL) games as the underlying event, this paper first revisits two of behavioral finance's major theories: overreaction and underreaction. These behavioral theories make somewhat conflicting predictions about investor activities, and I determine what evidence of each theory exists in NFL gambling markets. The paper then focuses on two hypothesized anomalies that are more specific to NFL wagering. These contradictions to the weak and semi-strong form of the efficient markets hypothesis in the NFL market I label and normal curve Implication ignorance involves the mispricing of contracts that depend on multiple distinct events. This analysis examines the prices of parlay contracts with an eye toward the behavioral work of Tversky and Kahneman (1974). Improper normal curve pricing investigates the prices of different point spread contracts in the context of Stern's (1991) normal curve approximation of the margin of victory distribution. Though no evidence of overreaction or underreaction in NFL trading is found, there is strong evidence of implication ignorance and marginal evidence of improper normal curve pricing.