Investor sentiment and cross-sectional stock returns.

This thesis consists of three essays on investor sentiment and the cross-sections of stock returns. The first essay extends Deling, Shieifer abd Waldmans (1990) noise trader risk module into a module with multiple risky assets to show the asymmetric effect of sentiment in the cross-section. Guided by our module, we also find that the effect of investor sentiment can be decomposed into long and short run components. The empirical tests in the first essay of the thesis present a negative relationship between long-run sentiment component and subsequent stock returns and a positive association between the short run sentiment and contemporaneous stock returns. The second essay explores a previously unexamined sentiment channel through which technical analysis can add value. We construct a daily market TA sentiment indicator from a spectrum of commonly used technical trading strategies. We find that this indicator significantly correlates with other popular sentiment measures. An increase in TA sentiment indicator is accompanied by high contemporaneous returns and predicts high near-term returns, low subsequent returns and high crash risk in the cross-section. We also design trading strategies to explore the profitability of our new TA sentiment indicator. Our trading strategies generate remarkable and robust profits. The third essay focusses on exploring the profitability of trading strategies based on Implied Volatility indicator (VIX) from the sentiment perspective. Our trading strategies involve holding sentiment-prone stocks when VIX is low and sentiment-immune stocks when VIX is high. The shifting asset allocation strategies are based on Abreu and Brunnermeiers (2003) delayed arbitrage theory and the asymmetric effect of investor sentiment in the cross-section. We find sentiment-prone stock have larger one-day forward retunes following high sentiment and vice versa. Our trading strategies generate substantial higher returns that benchmark portfolios, and the excess returns are not subsumed by well-known risk factors or transaction costs.