Effectiveness of and bp-neural Network for Overnight Stock Price Forecasting

Abstract

In view of the empirical problem of introducing machine learning into stock price prediction and the unique condition of Chinese stock market represented by the limitation of rise and fall, In our research, we take CSI300 component stock as samples, using BP neural network to predict stock price, introduce the factor representing the Piece Limit System into the analysis, and demonstrate the effectiveness of Chinese A-share market and the feasibility of neural network to forecast stock price. The study find that the BP neural network further improves the accuracy of stock price prediction with the function of logistic regression, and has a better AUC performance in the sample interval. Our research optimizes the machine learning process to provide more efficient empirical evidence of machine learning in Chinese stock market prediction in recent years.