Predicting stock price based on BP neural network model

Abstract

The increase in the stock will affect the development of the economy, so the description and forecast of the capital are fundamental. Therefore, this paper first analyzes the historical data of Everbright Bank stock (601818) and explains the stock increase, the share affected by the exchange rate, and the index correlation analysis. Then a prediction model based on BP neural network is established. BP is used to train stock descriptors to predict stock prices for the first 30 trading days in early 2019. Combined with ARE and MSE for error analysis, the average relative error is 3.6%, and the mean-variance is 0.041, indicating that the prediction effect of the model is good. To make a specific contribution to the study of stock price prediction.