Research on the Construction and Validity of Stock Market Investor Sentiment Index based on Stock Bar Comments

Abstract

In the present day, the internet has become the primary place for many retail investors in China to invest, and stock forums are often the primary choice for investors to exchange and discuss. According to behavioural finance, irrational investor decisions are often the source of instability in the stock market. Therefore, the study of investor sentiment has significant implications for the optimisation of investor decisions, the design of market surveillance mechanisms and the study of behavioural finance theory. This paper combines empirical research as well as theoretical analysis, a comprehensive analysis of the raw data extracted based on Eastern Fortune stock bar comments to establish the corresponding sentiment index, and will also give an analysis and conclusion on the validity verification of the sentiment index. The research includes the following two parts: (1) Using text mining and web crawlers, text-based data and numerical data are extracted, and data mining is completed after comprehensive analysis and screening, and text sentiment analysis is used to complete the establishment of the sentiment index for investors. (2) The full-day, trading day and overnight investor sentiment indices were compared with the SSE Composite Stock Index as the research object through correlation analysis with the mainstream six sentiment proxy variables to prove their significant correlation, and then the KNN regression based on simulated annealing and the LightGBM regression model based on genetic algorithm were used to verify the prediction of investor sentiment on the stock price movement of the SSE Composite Index respectively The empirical validity of the regression models is verified.In the end, this article will propose tailored policy optimization suggestions for both investors and market regulatory agencies, in accordance with the results of the analysis.