Are Crowds Wise When Predicting Against Point Spreads?
Abstract (Via Kellogg)
Point spread betting markets are considered an important example of crowd wisdom, because point spreads are accurate and are believed to reflect the ―crowd’s‖ predictions of sporting events. However, a season-long experiment found that a crowd of football bettors was systematically biased and performed poorly when predicting which team would win against a point spread. Moreover, the crowd’s biases worsened over time. However, when the crowd was instead asked to predict game outcomes by estimating point differentials, its predictions were unbiased and wiser. Thus, the same ―crowd‖ of bettors can emerge wise or unwise, depending on how predictions are elicited.
Introduction (Via Kellogg)
The wisdom-of-crowds hypothesis predicts that the independent judgments of a crowd of individuals (as measured by any form of central tendency) will be relatively accurate, even when most of the individuals in the crowd are ignorant and error-prone (James Surowiecki 2004). Examples abound (Surowiecki 2004; Ilan Yaniv 2004; Cass R. Sunstein 2006). Irving Lorge et al. (1958) found that students’ average estimate of the temperature of a classroom was only 0.4 degrees from accuracy, a result that was better than 80% of the individuals’ judgments. Jack L. Treynor (1987) asked 56 students to estimate the number of jelly beans in a jar. The average guess was 871, very close to the true number of 850, and better than 98% of the students’ individual guesses. And, Francis Galton (1907) reported the results of a regional fair competition that required people to estimate the weight of an ox. The average estimate was 1,197, just one pound away from the 1,198-pound ox’s true weight!
The practical implications of the wisdom-of-crowds hypothesis are tremendous. First, it suggests that decisions made by majority rule (or by averaging opinions) will often outperform decisions made by single judges or experts (Richard E. Larrick and Jack B. Soll 2006) or decisions made by group discussion (Sunstein 2006). Second, it suggests that decisions made by majority rule (or by averaging opinions) will often be accurate in an absolute sense, an implication that partially accounts for the rapidly increasing use of information markets to forecast events and to inform policy decisions (Robert W. Hahn and Paul C. Tetlock 2006; Teck-Hua Ho and Kay-Yut Chen 2007). Indeed, as detailed below, crowd wisdom has been implicated as a cause of market efficiency1 (Treynor 1987; Surowiecki 2004).
Given the wide-ranging implications of the wisdom-of-crowds hypothesis, it is important to know whether, and when, crowds will make wise decisions. In this paper, we test the wisdom-of-crowds hypothesis in a sports gambling context that features efficient prices that are widely believed to reflect crowd wisdom. We investigate whether crowds are typically wise in this context, whether crowd wisdom increases with feedback over time, and whether crowd wisdom depends on how predictions are elicited.