Noise dressing of the correlation matrix of factor models

Fabrizio Lillo, Rosario N. Mantegna

Published 2003-05-23Version 1

We study the spectral density of factor models of multivariate time series. By making use of the Random Matrix Theory we analytically quantify the effect of noise dressing on the spectral density due to the finiteness of the sample. We consider a broad range of models ranging from one factor models in time and frequency domain to hierarchical multifactor models.