Zooming into market states

TL;DR

This paper investigates the stability and dynamics of market correlation structures in the Nasdaq index over 22 years, revealing a relationship between average correlation and its fluctuations, especially during crises.

Contribution

It introduces a random matrix approach to model non-stationarity in market correlations, providing a new parameter to quantify correlation stability and dynamics.

Findings

Correlation structures cluster into distinct market states.

Correlation fluctuations are highest during market crises.

A single parameter effectively characterizes correlation stability.

Abstract

We analyze the daily stock data of the Nasdaq Composite index in the 22-year period 1992-2013 and identify market states as clusters of correlation matrices with similar correlation structures. We investigate the stability of the correlation structure of each state by estimating the statistical fluctuations of correlations due to their non-stationarity. Our study is based on a random matrix approach recently introduced to model the non-stationarity of correlations by an ensemble of random matrices. This approach reduces the complexity of the correlated market to a single parameter which characterizes the fluctuations of the correlations and can be determined directly from the empirical return distributions. This parameter provides an insight into the stability of the correlation structure of each market state as well as into the correlation structure dynamics in the whole observation period. The analysis reveals an intriguing relationship between average correlation and correlation fluctuations. The strongest fluctuations occur during periods of high average correlation which is the case particularly in times of crisis.