Bayesian stochastic volatility models for high-frequency data

TL;DR

This paper introduces a Bayesian stochastic volatility model tailored for high-frequency stock data that incorporates microstructure noise, providing accurate estimates of latent volatility across various sampling intervals.

Contribution

It develops a coherent Bayesian framework with a Markov chain Monte Carlo algorithm for high-frequency data that directly models microstructure noise and latent volatility.

Findings

Accurately estimates latent volatility from high-frequency data.

Handles microstructure noise effectively in a state-space model.

Provides well-calibrated volatility estimates in empirical tests.

Abstract

We formulate a discrete-time Bayesian stochastic volatility model for high-frequency stock-market data that directly accounts for microstructure noise, and outline a Markov chain Monte Carlo algorithm for parameter estimation. The methods described in this paper are designed to be coherent across all sampling timescales, with the goal of estimating the latent log-volatility signal from data collected at arbitrarily short sampling periods. In keeping with this goal, we carefully develop a method for eliciting priors. The empirical results derived from both simulated and real data show that directly accounting for microstructure in a state-space formulation allows for well-calibrated estimates of the log-volatility process driving prices.