This paper studies a financial market populated by adaptive traders. Learning is modeled following Camerer and Ho (1999). A call market and a Walrasian tatonnement are compared in an environment in which both institutions have the same Nash and competitive equilibrium outcomes. When traders learn via a belief-based model, equilibrium is discovered in both types of markets. In contrast, when traders learn via a reinforcement-based model, convergence to equilibrium is achieved in the Walrasian tatonnement but not in the call market. This paper suggests that market mechanisms can be designed to foster traders' learning of equilibrium strategies.