Identifying and Pricing Adverse Selection Risk with VPIN

Abstract

We perform the first large-sample estimation of the Volume Synchronized Probability of Informed Trading (VPIN) measure on the NYSE TAQ universe, enabling us to test the validity of VPIN with high statistical power and to do traditional asset pricing tests of informed trading. Informed trading measured by VPIN is priced, and is not explained by firm characteristics such as volume, volatility, or liquidity, supporting the validity of the measure. Additionally, we create a novel signed version of VPIN to identify the direction of informed trades. A portfolio long low-VPIN stocks and short high-VPIN ones delivers a monthly five-factor alpha of .18%, which rises to .29% when using signed VPIN. A trading strategy following this signed VPIN factor delivers an annualized five-factor BHAR of 11.45%. We further document a reversal in stock performance in portfolio sorts on signed VPIN, the incorporation of which into a trading strategy improves performance to an annualized BHAR of 17.34%.