Bayesian Estimation for High-Frequency Volatility Models in a Time Deformed Framework

This article addresses problems associated with the estimation of parameters for models used in modeling high-frequency financial volatility. Decision-making involving the allocation of resources within financial markets depends heavily on risk measures, and the reliability of models that provide a volatility analysis is crucial. With a parametric framework, the Stochastic Volatility model is capable of giving answers concerning the volatility evolution, estimated through Bayesian methods. With the availability of intraday data, there is the extension of models for coping with observed characteristics of financial returns. One of such extensions is the two factors model. The parameter estimation uncertainty can be so significant that it may impair any reasonable interpretation. One reason is the insufficient information that a unique series of prices can provide to separate the effects of an enormous variety of parameters added to models. The proposed approach consists of harnessing new information sources capable of improving estimation processes. The strategy is to consider a different time frame. By analyzing the problem within a time-deformed framework, or “operational” time, further information may be gathered, which can allow models to be more efficiently estimated. Extensions use price, volume, and duration variables in the models’ redefinition.