Quantitative Analysis Of Financial News And Stock Price Fluctuations

The stock market has always been an area of great interest to many investors and scholars.The beautiful fluctuations in stock prices make no mathematics a mystery;for many investors,if they can predict the future trend of the stock market,they can change their investment decisions in advance,thus helping them to obtain excess profits.Although there are many factors affecting the stock market trend,ranging from national policy changes,external market environment and natural disasters,to financial indicators such as the company's own operating conditions and debt capacity,these factors are ultimately transmitted in the form of information.Investors can obtain this information through the media and other channels and make corresponding investment decisions based on this information,thus causing fluctuations in the stock market.We often see that the state or listed companies publish some important news.Once this news is released,it will spread quickly on the online media.Regardless of whether the information carried in the news is good or bad,it will affect the investor's investment decisions intentionally or unintentionally,resulting in stock market volatility.Moreover,the volatility of the stock market will also be transmitted to the stock market participants through the news media in the form of information,and participants in the stock market will change their investment decisions based on the information obtained.Especially with the popularity of the Internet,more people use the Internet to obtain information,which makes the impact of Internet financial news more powerful and has a wider range of influence.Therefore,it is particularly important to study the relationship between Internet financial news and stock price volatility,and provide some reasonable suggestions for the trading activities of stock market participants.This paper is based on text processing technology,regression analysis,support vector machine and random forest algorithm.The constituents of the above 50 index are used as the research object to study the relationship between financial news and stock price fluctuation.This article uses python to write the web crawler code,and obtains the stock price historical data of the SSE 50 index constituent stocks published on the Sina Finance website from January 1,2017 to December 31,2018,and the Netease Financial website SSE 50 index constituents..In order to convert unstructured data into structured data,firstly use text segmentation quantification technology to quantify the obtained financial news texts and extract the top 100 keywords after the word segmentation.Secondly,in order to reduce the sparseness of dimensions and matrices,it is convenient for subsequent analysis.The news keywords are divided into seven categories of words,and the weight value of each type of words in each financial news is calculated,and the irrelevant words are removed to obtain the final financial News quantitative data.Finally,the quantitative data of the financial news and the historical data of the stock price are collated,the data is organized and data for modeling is acquired.In order to explore the relationship between financial news and stock price fluctuation and the relation between different categories of words and the closeness of stock price fluctuation,this paper presents the weights of six kinds of words and the rise in stock price and the announcement of financial news Fall after.The direction of the ups and downs is the target variable.A model was established using regression analysis,support vector machine and random forest algorithm.Taking into account that the stock price history data itself is time series,stock price historical data coefficients are added to the model process to accurately quantify the relationship between financial news and stock price fluctuation and to improve the accuracy of the model I will.Empirical analysis leads to the following conclusions.There is a complex nonlinear relationship between financial news and stock price,and this relationship will be the largest on the first day after the news release.Three words in Internet financial news text such as emotional derogatory words,aggressive corporate status and emotional derogatory words are closely related to stock price fluctuations.Stock price fluctuations are not that big.The model established by adding stock historical data works better than all models using only quantitative data of financial news,which not only improves classification accuracy of classifiers but also squares regression equipment Also reduce the mean square root error.Considering the T + 1 transaction policy of the domestic stock market,the closing price for the day is selected as the research level.The best estimate time for the stock price of the Internet financial news text is the closing price of the first trading day after the press release.In the simulation of investment with a simple investment trading strategy,it was confirmed that when using the prediction result of the closing price of the first trading day after the news release of stock investment,certain advantages can be obtained.