Research On The Influence Of Financial News On Stock Market Investment Decision Based On Deep Learning

Volatility and prediction of stock prices in the stock market has always been the concern of investors and one of the hot spots in academic research.Past research mainly used trading data to do stock price forecasting and did not get good results.With the emergence of deep learning and text mining techno logy,this paper makes more research on stock market price volatility.This paper uses behavioral finance,text mining and deep learning related knowledge and theory of stock market news data sources.First of all,this paper combs the literature review about financ ial news and stock market,and points out that the methods used in previous studies can not dig out the connotation information of financ ial news well enough to explain the stock market in the prediction or analysis of financ ial news.This paper uses Python to write a web crawler crawl the news headlines from January 1,2016 to December 31,2017 in the Wall Street news letter and the corresponding release time as the initial sample data of financ ial news.At the same time from the wind database to obtain the January 4,2016 to December 29,2017 transaction data,includ ing the opening,the highest price,the lowest price,trading volume,prices.Second,the financial news data from the non-trading day and the remaining news data as research sample data,hereinafter called Financ ial News sample data.First,the annual statist ics of financ ial news text data and stop using a part of high-frequency words without the actual meaning of the word to obtain the first 100 words per year Baidu Index,in the sample period,the yield of the Shanghai and Shenzhen 300 index is the dependent variab le,the index of 100 words is the independent variable,and the random forest algorithm is used to return the The important words in the sample period are obtained,and the results show that the important words coincide with the hot spots in the sample period,which proves that the financ ial news can really affect the investors ' trading decisions.Secondly,we label the sample data of financ ial news according to the fluctuation,if the increase is greater than 0 with the sample data of financial news at the same time,it is 1,otherwise it is-1.The data that can be used for deep learning training is obtained.This paper uses word 2vec to train the samp le data of financ ial news,and then uses long short term memory neural network(long short term me mory)to train the data to get a stable output model,and according to the trained model to predict the rise and fall of Shanghai and Shenzhen 300;and compares the prediction results under the same conditions using only financ ial time series data?The results show that the prediction effect of the model is improved after adding financial news text data.Convolutional neural network performance under the same conditions is compared.The results show that it is superior to convolutional neural network in financial time series data,financ ial news text data and financial news text data combined with financial time series data.