Neural Network Investment Strategy Algorithm

Abstract

Abstract. Human brain is a highly complex and nonlinear parallel information processing system. The artificial neural network built by imitating the working mechanism of human brain has strong nonlinear input-output mapping ability. Based on the previous data, the future price data of gold and bitcoin are predicted through the construction and training of neural network. In the short-term forecasting model, the multi-objective programming with minimum net benefit and maximum risk is adopted. In the process of constructing the long-term prediction model, we conducted machine learning on the three indexes LINES, KDJ, MA, and obtained the optimization model of single index decision. In addition, we test the optimality of the model. We also study the short-term decision model. Finally, the optimal solution is obtained by randomly changing the weights of the three indexes.