A Stochastic Feedback Model for Volatility

TL;DR

This paper introduces a simple stochastic feedback model for volatility that captures key empirical properties of financial time series, such as fat tails, autocorrelations, and multifractality, better than standard models.

Contribution

It proposes a parsimonious volatility model with only two parameters, grounded in a framework involving estimation error of an exogenous Poisson rate, improving empirical fit.

Findings

Model reproduces fat tails and autocorrelation in absolute returns.

Fits empirical data better than standard volatility models.

Grounded in a framework involving estimation error of Poisson rate.

Abstract

Financial time series exhibit a number of interesting properties that are difficult to explain with simple models. These properties include fat-tails in the distribution of price fluctuations (or returns) that are slowly removed at longer timescales, strong autocorrelations in absolute returns but zero autocorrelation in returns themselves, and multifractal scaling. Although the underlying cause of these features is unknown, there is growing evidence they originate in the behavior of volatility, i.e., in the behavior of the magnitude of price fluctuations. In this paper, we posit a feedback mechanism for volatility that closely reproduces the non-trivial properties of empirical prices. The model is parsimonious, contains only two parameters that are easily estimated, fits empirical data better than standard models, and can be grounded in a straightforward framework where volatility fluctuations are driven by the estimation error of an exogenous Poisson rate.