Investigating the Role of Sentiment Measures in Explaining Anomalies

Abstract

This thesis provides a comprehensive examination of the role of investor sentiment in explaining market anomalies that challenge both weak‑form and semi‑strong‑form market efficiency. It integrates twelve sentiment measures—spanning survey‑based indices, composite market indicators, and novel textual‑based metrics derived from corporate earnings calls and financial reports—alongside seventeen well‑documented anomalies. Using monthly data, a reduced‑form vector autoregression (VAR) framework, complemented by Granger‑causality tests, impulse response analysis, and variance decomposition, uncovers the dynamic causal relationships between sentiment shocks and anomalous return patterns across contrarian, momentum, reversal, and many other factor strategies over short and long horizons. Textual‑based sentiment indices deliver the strongest explanatory power, especially for weak-form market anomalies, outperforming traditional survey and composite indices. These findings advance the behavioral asset‑pricing literature by quantifying the added value of soft‑information extraction method. They also underscore the practical implications for integrating sentiment metrics into forecasting models, risk assessments, and trading strategies. By considering multiple sentiment dimensions and anomaly categories, this study fills a critical gap in behavioral finance and offers actionable insights for academics and investment professionals alike.