The weekly cycle of investor sentiment and the holiday effect-- An empirical study of Chinese stock market based on natural language processing

Abstract

Investor sentiment is an important factor that affects stock prices, stock market returns, and asset pricing. However, the fluctuation patterns and factors influencing investor sentiment have received less attention from scholars. This study uses text messages from stock investors’ social networks and natural language processing techniques to reveal sentiment fluctuation laws of stock market investors. An investor confidence index (ICI) is constructed by quantifying sentiment in investor messages on social networks. By taking this index as a proxy for sentiment, we measure the candidate fluctuation periods of investor sentiment using a Fourier transform. The significance test then determines the significant cycle of investor sentiment within seven days. Based on this, cluster analysis further reveals that investor sentiment in the 7-day cycle has a 5 + 2 cycle of variability. That is, from Monday to Friday, investor sentiment is disturbed by stock market sentiment showing profit-seeking and risk-averse preferences, while during the weekend holiday, stock market disturbance to investor sentiment becomes lower, investor sentiment is substantially higher, and volatility is narrowed, showing a typical holiday effect. The analysis also shows that the recurring cycle of 5-day trading days and 2-day holidays is a direct exogenous factor contributing to the 7-day cycle of investor sentiment. This study provides a new perspective for studying “investor sentiment,” “day of the week effect,” and “behavioral finance.”