This dissertation consists of two chapters on finance and experimental economics.

The first chapter studies the dynamic portfolio optimization problem with reinforcement learning. I evaluate several algorithms on simulated data to document their convergence properties and sample efficiencies. I also apply a state-of-the-art algorithm on two empirical problems and show that they outperform other traditional strategies.

The second chapter studies alternating bargaining games, by proposing an ultimatum game with uncertainties. I model the two-stage game as a screening game that incorporates the social factor of fairness, and run experiments to analyze how participants behave in response to bargaining power shift.