Suppose it’s a home match for the Golden State Warriors, and Steph Curry demonstrates his prowess early on, sinking consecutive three-pointers within minutes of the first quarter starting. The fans at the Chase Center are impressed, and the impact is immediately noticeable in the betting markets as the odds shift to favour the Warriors.
However, the game is far from over. The opposing team mounts a comeback, and with only 10 seconds remaining, the Warriors find themselves trailing by two points after missing a crucial shot. Logic dictates that the betting odds should now favour the away team, considering their advantageous position. Yet, the odds remain unexpectedly stable.
Eben Lazarus, an assistant professor of finance at UC Berkeley’s Haas School of Business, explains that according to historical NBA game data, a team in possession of the ball with a two-point lead and only 10 seconds left on the clock has over 90% chance of winning. Nevertheless, betting markets seem to overly emphasise early-game events, such as initial baskets, while downplaying significant late-game developments.
This phenomenon of misinterpreting new information isn’t confined to sports betting — it’s prevalent across financial markets, too, as revealed in a study published in the Quarterly Journal of Economics. Lazarus, along with Ned Augenblick from UC Berkeley Haas and Michael Thaler of University College London, conducted three experimental studies and analysed millions of betting transactions and options contract prices. Their findings consistently showed that people overreact to relatively insignificant information and underreact to more crucial data.
Lazarus comments on the standard difficulty in gauging the true significance of new information, regardless of whether its impact is positive or negative. This pattern of skewed reactions to information has been observed universally, much to the researchers’ surprise, given the high stakes in both betting and financial market environments.
Their research builds upon decades of behavioural psychology and economics studies, exploring how people update their beliefs in response to new information. This includes seminal works such as a 1966 paper suggesting people are overly conservative in revising their beliefs and a 1992 study by Dale Griffin and Amos Tversky that indicated a tendency to focus excessively on dramatic information while neglecting its reliability.
Recent studies further show that people often err systematically due to miscalculations in probability and a propensity to choose a middle-ground option when unsure. The paper also ties into broader observations of financial markets occasionally overreacting or underreacting to news.
“We believe our framework simplifies the analysis of numerous scenarios in financial markets and real-world settings,” Lazarus asserts. He explains that while we constantly absorb new information—from election polls favouring a particular candidate to feedback from a superior—accurately assessing its significance often eludes us, leading to default middle-ground responses.
In one innovative experiment involving 500 NBA fans, the research team simulated parts of basketball games, asking participants to estimate win probabilities after each sequence of events. Despite recognising the greater importance of late-game events, participants significantly overvalued early baskets and undervalued those nearer the end of the game.
Further tests using real-world sports betting data from Betfair and option price quotes from the Chicago Board Options Exchange confirmed the same pattern: early events in a game disproportionately affect betting odds despite their limited actual predictive value, while crucial late-game events fail to sway the market as much as they should.
Lazarus cautions that recognising these market patterns only partially eliminates risk. He advises considering how much weight to give different pieces of information, especially in ambiguous situations. He concludes with a personal anecdote about overreacting to a negative interaction with a boss, suggesting that people often benefit from taking a step back to evaluate the real significance of such events.
More information: Eben Lazarus et al, Overinference from Weak Signals and Underinference from Strong Signals, The Quarterly Journal of Economics. DOI: 10.1093/qje/qjae032
Journal information: The Quarterly Journal of Economics Provided by University of California – Berkeley Haas School of Business