Research On Dynamic Copula-wavelet SVM Model Based On Stock Market Correlation And Investor Sentiment

The stock market is an important part of our capital market and it plays an significant role in optimizing the allocation of social resources and promoting economic development.With the gradual improvement of people’s material living standards,their investment awareness continue to strengthen.The stock market attracts many investors with its strong speculative nature and it promotes the socialist market economy with Chinese characteristics sustainable and healthy development.In order to improve the accuracy of stock prediction,this study fully considers the influence of stock market correlation,investor sentiment,random fluctuations and other factors,proposes a dynamic Copula-wavelet support vector regression model based on stock market correlation and investor sentiment.And this paper takes Shanghai 50 shares as an example to carry on the application research.The study consists of four main parts.The first part is the stock market correlation research.In statistics,correlation coefficient is usually used to measure the correlation between variables.But the traditional correlation coefficient is always limited to many constraints,so it is unable to systematically and comprehensively describe the dependent structure of variables.Since Copula function can well describe the linear and nonlinear,symmetric and asymmetric correlation between variables,and analyze the correlation structure between variables,so this paper uses the dynamic time-varying Copula function to measure the dependent structure between logarithmic returns of different stocks.Stocks with strong correlation with target stocks are selected by dynamic upper and lower tail correlation coefficients and form the stock correlation set.The second part is investor sentiment analysis.Traditional stock forecasting researches focus on index selection and model optimization.They ignore the importance of investor sentiment.However,as the cornerstone of the whole ecological operation,investors play an important role in the healthy development of the capital market.In the current stock market,small and medium-sized investors play an important role.Most of them do not have relevant professional knowledge and they are easy to follow the trend.Therefore,investor sentiment should become one of the key factors in the study of stock price trend.This paper uses web crawler technology to obtain web comments about target stocks from Guba website and constructs investor sentiment index by text analysis method.The third part is wavelet support vector regression model.The classical Gaussian kernel support vector regression model is not ideal in the noisy stock market.This paper introduces wavelet theory to reduce the influence of noise,that is,introduces Morlet wavelet kernel function into the support vector regression model.Then combined with the closing price of each stock in the stock correlation set and the investor sentiment index of target stock to predict the closing price of the target stock the next day by the Morlet wavelet support vector regression model.The last part is application research.This paper chooses the data of Shanghai Pudong Development Bank as the target stock and Shanghai 50 shares except China Telecom and China International Capital Corporation from June 22,2020 to March 15,2022 for research.This paper tests the impact on the prediction accuracy of the model of stock correlation set,investor sentiment and wavelet kernel function.It is proved that this model can effectively improve the accuracy and stability of stock price trend prediction.The main contribution of this paper is reflected in three aspects.First,this paper fully considers the influence of stock market correlation on stock price trend prediction and measures the tail correlation between logarithmic returns of stocks by the dynamic Copula model.And then this paper constructs the stock correlation set by the tail correlation.Second,this paper constructs a specific financial comment sentiment dictionary by Python.To improve the prediction accuracy of the model,this paper combines the Hownet sentiment dictionary and the most commonly used Chinese word segmentation database jieba to conduct sentiment analysis and integrated the analysis results with the sentiment analysis results of Rost Cm6 software which developed by Wuhan University to establish an exploratory investor sentiment index.Thirdly,this paper gives full play to the noise reduction function of wavelet kernel function and realizes the real combination of wavelet theory and support vector regression model,and improves the generalization ability and prediction performance of the model.Model combination is one of the focuses of current research on stock price prediction.In this paper,dynamic Copula model,text analysis method and Morlet wavelet kernel support vector regression are combined for the first time.The model in this paper effectively improves the accuracy of stock price trend prediction.It is an exploration and innovation of combination model.