LAUSR

One crucial task of option price modeling is to estimate latent state variables. This paper emphasizes the importance of incorporating option implied information to update latent state variables and sheds light on numerical developments to alleviate the cumbersome estimation process in option valuation. We propose a simple option-implied approximation to obtain the latent state variable and investigate its performance in option pricing. Specifically, we directly filter conditional variance from option implied volatilities (option-implied filtering) and compare its performance to that of a futures-based filtering technique and that of an option-based filtering technique with the Brownian Bridge process. Using a GARCH type discrete-time option pricing model and the crude oil option data, we demonstrate that the option-implied filtering technique significantly improves model fit and estimation efficiency, both in-sample and out-of-sample.