Relative Investor Sentiment Measurement

Abstract

This paper proposes a new metric to gauge investor sentiment using a relative valuation method. We combine investor behavioral finance traits and option-implied standard deviations under both the real-world probability (P) valued most in the view of uninformed investors and the risk-neutral space (Q) adopted when there exists no cognitive error. Given that investor sentiment can be thought of as risk-taking by the uninformed exceeding their informed peers, we postulate that the differences between the variance, skewness, and kurtosis of P and Q measures for investors with various behavioral traits matter. We hence construct our investor sentiment proxy by summing these differentials of variance, skewness, and kurtosis in weighted forms. It is documented that such relative investor sentiment metric exhibits economically and statistically strong return predictability for momentum portfolios. Our findings contribute to the extant literature by (1) complementing the Baker-Wurgler market-based investor sentiment index from the theoretical perspective, (2) modeling investor sentiment via utilizing the informational content of options prices, and (3) supporting the Barberis-Shleifer-Vishny definition of investor sentiment to be differences in financial market participant behavior.