| Summary: | 碩士 === 國立交通大學 === 財務金融研究所 === 107 === In this paper, we use overnight return (CTO) as an indicator of investor sentiment, and intraday return (OTC) as an indicator of investor sentiment recovery to predict future stock price changes. Firstly, we study the effects of different industries, company characteristics and business cycles on the variables in the regression. At the same time, we divide the variables in the regression into five groups from low to high. We calculate the abnormal return (AR) of each group in the next 1 month, 3 months, 6 months, 9 months and 12 months. We found that the portfolio of Intercept, R, and ORVOLD in the regression model can capture significant positive returns for the next 3-12 months. We then found that arbitrage combinations can still be achieved with different company characteristics, and that different company characteristics and economic cycles can affect the ability of variables to capture abnormal returns. Subsequently, we added the abnormal return (AR) as a factor to the FF5 factor model. We found that the AR9 and AR12 captured by Intercept can improve the explanatory power of the five-factor model and cannot be explained by the five factors and Qfactor.
Keywords: Overnight Returns, Investor Sentiment, Information Shock, Abnormal Return,Factor Model
|
|---|
