The Fantasy of Expertise: Professional and Algorithmic Opacity in a Prediction Market

Saturday, June 25, 2016: 4:15 PM-5:45 PM
247 Dwinelle (Dwinelle Hall)
Jeff Gordon, U.C. Berkeley, Berkeley, CA
Do amateur investors and gamblers place more confidence in predictions generated with the help of statistical modeling algorithms? In financial markets and analogous gambling arenas, statistical modeling is generally considered a tool by which experts ply their advantage. I argue that amateur market participants view such tools as leveling forces and, with ballyhooed technology in hand, overestimate their chances of out-predicting the experts. I analyze 2,091 comments on online message boards posted by participants in Daily Fantasy Sports contests, a sports gambling game that bears resemblance to stock market day trading. In these contests, 1% of players win 91% of profits. Many players perceive the biggest winners to rely on a mastery of statistical modeling and therefore seek to emulate the use of such models. But as Burrell (forthcoming) demonstrates, algorithms and models are “opaque” to human (and especially non-expert) interpretation. This opacity engenders a willing naïveté for the gamblers, who ascribe more predictive power to their tools than they would to their own, unassisted reasoning. This finding bears on the social circumstances of flash crashes and other unintended consequences of algorithmic opacity in financial markets.