A Stochastic Price Duration Model for Estimating High-Frequency Volatility

Abstract We propose a stochastic price duration model to estimate high-frequency volatility. A price duration is directly linked to volatility from the passage time theory for Brownian motions, and it possesses several advantages over returns for estimating volatility. We employ price durations in a parametric model that directly specifies stochastic volatility dynamics. Our approach allows us to estimate intraday spot volatility and our empirical results suggest the presence of important intraday volatility dynamics. We conduct an extensive integrated variance forecast comparison, which demonstrates the superior performance of our proposed models compared with other duration-based or return-based estimators.