Research On Stock Price Trend Prediction Based On Deep Learning

Stock is an important part of capital market construction.The change of stock market is of great significance to social and economic development,investment institutions and investors.Therefore,the prediction of stock price trend has always been the focus of scholars’ research.However,the change of stock price is affected by many factors,and its price change is characterized by non-linearity and high noise.Traditional stock forecasting methods rely on manual data processing,which has strong subjectivity and low forecasting efficiency.With the rocketing progress of deep learning algorithm,more and more scholars apply deep learning algorithm to the research of stock price prediction in recent years.Based on the existing deep learning methods,this paper puts forward two improved algorithm models to reduce the prediction error and improve the accuracy of prediction.Firstly,this paper discusses the research background and current situation of stock price prediction,and describes the structural arrangement and innovation of this paper in detail.Secondly,the definition and related knowledge of the method involved in stock price prediction are introduced,and its working principle and process are described,mainly including decision tree,random forest(RF),cyclic neural network(RNN)and long-term and short-term memory network(LSTM).Then it introduces the preprocessing work before stock price prediction,including stock data selection,feature set construction and data normalization.Due to the large amount and complexity of stock data,the prediction accuracy of previous stock prediction models is low and the training is complex.This paper proposes an RF-LSTM prediction method,which combines random forest with long-term and short-term memory network algorithm to predict the closing price of stocks.The model is established by using the real stock market data,and the parameters of the prediction model are optimized.The experimental results show that the prediction model based on RF-LSTM proposed in this paper can predict the stock price more accurately,and provide a reference for controlling the price trend of the stock market.Then,a stock trend prediction model CNN-GRU is established by using convolutional neural network(CNN)and gated recurrent unit(GRU)learning algorithm to predict the future rise and fall of stocks.The trading data of Shanghai Stock Exchange and Shanghai Pudong Development Bank in 524 trading days were selected as the experimental object.CNN,LSTM,GRU and recurrent neural network(RNN)are set as comparative experiments.The experimental results show that the CNN-GRU stock trend prediction model established in this paper has good prediction performance on the experimental data set.There are 29 figures,10 tables and 55 references in this paper.