Research On Stock Price Forecasting Based On Emotion Tendency Of Financial News

With the stable development of China’s economy,the improvement of people’s living standard and the increasing of disposable income,how to invest idle funds rationally has become a topic attracting more and more people’s focus.It is why that stock as a high-risk and high-return investment mode has received widespread attention.Therefore,there is no doubt that how to avoid the stock market risk effectively in order to improve the return on investment are being a question to be discussed and studied constantly by the financial industry.In recent years,the key point of studies on stock forecasting has shifted from the traditional forecasting analysis based on the key indicators of stock to the forecasting analysis based on the combination of key indicators and basic information such as financial news gradually.Behavioral finance theory points out that texts such as financial news could influence investors’ decisions by affecting investors’ sentiment,leading to the change of stock price eventually.According to the theory,a lot of studies on stock forecasting based on news emotion tendency have emerged.However,the current studies in this field still have the following problems: 1)the analysis of news emotion tendency is mostly a kind of analysis based on the category of emotion tendency,but there are few analysis and applied studies carrying out from the fine-grained level of emotion tendency intensity;2)the studies based on the category of news emotion tendency cannot get the more accurate intensity of news emotion tendency,so they cannot effectively improve the accuracy of stock forecasting,especially in the aspect of forecasting the specific value of stock price.In order to solve the above problems,this paper designs a quantitative method of the intensity of financial news emotion tendency,and combines the quantitative index and some key indicators of stock.On this basis,the paper adopts the improved particle swarm algorithm to optimize the parameters of support vector regression model,eventually constructing a new model of stock price forecasting.The core work includes the following two parts:(1)A quantitative method of the intensity of news emotion tendency is designed,which is based on the weight improvement of emotion dictionary of financial news.Through combining the existing mature emotion dictionaries and the artificially selected emotion vocabularies in the financial field,this paper constructs an emotion dictionary for the financial field.On this basis,after fully considering the structural characteristics and emotion distribution of the financial news texts,the quantitative method of financial news emotion tendency is designed based on the weight improvement of lexical,location and semantic rules.Through the empirical analysis of financial news emotion tendency,the paper finds the superiority of the quantitative method of financial news emotion tendency designed in the paper.(2)A stock price forecasting model based on improved PSO-SVR is built.This paper fuses the index of financial news emotion tendency with the key indicators of stock extracted,and optimizes the parameters of support vector regression using improved particle swarm optimization algorithm,eventually constructing the forecasting model in the paper.The empirical result shows that the forecasting accuracy of the model built in the paper is higher than that of other comparison models.