On collective non-gaussian dependence patterns in high frequency financial data

TL;DR

This paper investigates non-Gaussian dependence patterns in high-frequency financial data, revealing phenomena like dependence-induced volatility smile and kurtosis reduction, explained through a multivariate t-Student distribution model.

Contribution

It introduces a model based on multivariate t-Student distribution to explain collective non-Gaussian dependence patterns in high-frequency financial data.

Findings

Dependence-induced volatility smile observed

Kurtosis reduction in intraday price increments

Multivariate t-Student model explains dependence effects

Abstract

The analysis of observed conditional distributions of both lagged and simultaneous intraday price increments of a basket of stocks reveals phenomena of dependence - induced volatility smile and kurtosis reduction. A model based on multivariate t-Student distribution shows that the observed effects are caused by colelctive non-gaussian dependence properties of financial time series.