Research On Prediction Of Stock Opening Price Based On TensorFlow Recurrent Neural Network

Stock is an important direct financing method for enterprises.It provides a way for individuals to increase their wealth.Stock is also an important tool for social wealth redistribution.However,high returns and high risks of stocks coexist.People are often attracted by high returns and ignore high risks.Predicting and analyzing the trend of stocks,so as to guide investors’ trading behavior,can effectively reduce risks and achieve high and stable investment returns.Although stock technical analysis and fundamental analysis have a certain predictive effect,but they are complicated and highly subjective.Quantitative analysis is becoming more and more popular by building models to predict stock trends.In recent years,with the rapid progress of computer technology,artificial intelligence has become a hot research field and breakthroughs have been made in deep learning.The models used in quantitative analysis have transitioned from linear to nonlinear.The prediction ability and reliability have been greatly improved.The robo-advisory services provided by financial institutions are gradually accepted by investors.This study uses TensorFlow machine learning platform,based on the traditional stock volume and price analysis theory,and uses BN and GRU to construct the opening price rise or fall and opening price prediction model for the next and next five trading days.The stock transaction data of Shanghai Composite Index,Maotai and China Construction Bank in recent ten years are used to verify the validity of the model.Compared with the traditional linear model,the model constructed by this method has good generalization ability and fitting ability.In particular,the prediction of opening price rise or fall for the next five trading days has a high accuracy rate and can be applied to the short-term arbitrage trading strategy of stocks.