Application of Deep Q-Network in Portfolio Management

Machine Learning algorithms and Neural Networks are widely applied to many different areas such as stock market prediction, facial recognition and automatic machine translation. This paper introduces a novel strategy based on the classic Deep Reinforcement Learning algorithm, Deep QNetwork, for stock market portfolio management. It is a type of deep neural network which is optimized by Q Learning. To adapt the Deep Q-Network for stock market production, we first discretize the action space so that portfolio management becomes a problem that Deep Q-Network can solve. Following this, we combine the Convolutional Neural Network and dueling Q-Net to enhance the recognition ability of the algorithm. We choose five low-relevant American stocks to test our model. It is found that the Deep Q-Network based strategy outperforms the ten other traditional strategies. The profit of Deep Q-Network algorithm is 30% more than the profit of other strategies. Moreover, the Sharpe ratio and Max Drawdown demonstrates that the risk of policy associated with Deep Q-Network is the lowest.