Portfolio Research Based on Neural Network and Multi-Objective Programming

摘要

In financial investment, risk and benefit coexist. How to balance the benefits and risks and find the optimal investment portfolio is a key issue to be considered by investors. In this research, BP neural network is used to predict the future return on equity (ROE) of asset; a multi-objective programming model of investment portfolio is established on the basis of Markowitz's portfolio investment theory; to select the optimal investment portfolio, a comparison is made on the benefit-risk ratio of the investment portfolio with the smallest risk and at different income levels; in addition, an empirical analysis is made on the basis of the quarterly ROE data of 5 stocks during 2002-2017.