As a collective phenomenon, the wisdom of crowds—the tendency for the average belief in a group to be more accurate than any single group member—offers no direct benefit to individuals. Financial markets, for example, demonstrate incredible efficiency even when most individual investors make poor decisions. Popular theoretical accounts hold that individuals must remain independent to preserve collective accuracy, a requirement that prevents social learning that might benefit individuals. However, formal models of social influence predict that properly structured information exchange allows individuals to learn from each other without undermining group accuracy. We develop an experimental study in which investors make financial forecasts before and after learning the beliefs of others in a structured social network, to explore the effects of networked collective intelligence on predictive accuracy and sentiment cascades in financial markets.