Research on the Interaction Between Investor Sentiment and Stock Market Returns‐‐Based on Random Forest Feature Selection Algorithm

10.6911/WSRJ.202206_8(6).0060

Yiwei Shao；Yihui Wu

Investor sentiment ； Stock market returns ； Random forest ； VAR model

World Scientific Research Journal

8卷6期（2022 / 06 / 01）

465 - 473

英文

The measurement of investor sentiment is a difficult and hot issue in behavioral finance, and the selection of indicators is the key and prerequisite for measurement. In this paper, investor sentiment indicators were screened according to the principles of availability, importance and relevance. Firstly, random forest was used to rank the importance of the indicators, and those of low importance were discarded. Secondly, after removing the components of the indicators that were not relevant to investor sentiment, principal component analysis was used to construct investor sentiment. Finally, a VAR model was constructed to analyze the dynamic correlation between changes in investor sentiment and changes in stock market returns. The following conclusions were drawn: high investors sentiment has a positive driving effect on stock market returns. At the same time, positive returns can inspire optimism in investors, which is inertial in the short term. After the third period, optimistic investor sentiment will gradually wane.