Emotional Reinforcement Learning for Portfolio Selection

Reinforcement learning algorithm has been successfully used in prediction and decision making [5,11]. The main contribution of this paper is to provide decision making using reinforcement learning approach to allocate resources optimally in stochastic conditions in a well known example; in the portfolio selection. The modern theories of portfolio selection consider some presumptions. But if they don’t hold, these methods are no longer efficient. So these days, some papers have been written by using the artificial intelligent methods. In this paper, appropriate emotional reinforcement signal is composed for portfolio selection. For this purpose, the reward signal is taken as the output of a linguistic fuzzy inference system with the return of portfolio and the risk of portfolio as inputs, then we implement the Q-learning neural network and we train this network with the proposed reward signal.