The Risk of Finance Words

Abstract

This paper proposes a dictionary tailored for volatility analysis in finance research. We investigate the comovement between corporate textual information and option-implied volatility, via robust multinomial inverse regression. The volatility dictionary contains vastly different words from the sentiment dictionary (Loughran and McDonald, 2011) and the colour dictionary (Garcia et al., 2023). The signals distilled from the volatility dictionary explain the cross-sectional variation in implied volatility dynamics, as well as the levels of implied and realized volatility. We find the volatility signals diminish within one day, which is much faster than the assimilation of the expected return signals.

Keywords: Textual Analysis, Option-Implied Volatility, Machine Learning, Risk Management, Information

JEL Classification: C53, G10, G14