Empirical Analysis Of Stock Price Based On Recurrent Neural Network

With the rapid development of economy and society,the people's disposable income is getting higher,the Engel coefficient is getting lower,and the capital investment activities are becoming more common.Stock investment,a high risk and income investment,has attracted more attention.In recent five years,with the deepening of supply side reform and financial system reform,China's stock market is becoming more mature.The size and volume of two major stock exchanges in Shanghai and Shenzhen increase exponentially.At present,the operation of stock market in China shows some new characteristics.It is necessary to explore the underlying rule of stock price and its change so as to explore the meaning and value of each stock's characteristic quantity.As the most popular algorithm in machine learning and data mining,neural network not only plays a prominent role in speech recognition and image classification,but also has obvious advantages in dealing with long time series prediction.It is noted that the research and prediction of stock price depend on the stock price in the past time,while the traditional neural network method can not solve this problem well.This paper adopts LSTM structure modeling of recurrent neural network and makes a research of September 1,2003-2017 year in December 29 th the gold stock price.The stock mainly selects the high price,low price,opening price,closing price,trading volume and turnover of the 6 variables.At first it will train and test the prediction in recurrent neural network structure.Then the loss function,prediction accuracy and root mean square error are statistically diagnosed,and our results are compared with the results of BP neural network structure.Finally,we will give evaluation and prediction analysis.The results obtained in this paper can provide investors with certain investment reference,and can also serve as a theoretical basis for decision-making of relevant government departments,so as to promote the healthy and development of the securities market better.