On the analysis of cross-correlations in South African market data

We report briefly on an application of random matrix theory to the analysis of SA financial market data (An Analysis of cross-correlations in South African financial market data, e- print cond-mat/0402389). Correlation matrices C are constructed from 10 years of daily data for stocks listed on the Johannesburg Stock Exchange from January 1993 to December 2002. Spectral properties of C are tested against random matrix predictions. We highlight some quantitative differences which arise when treating prices as existing only when measured, as opposed to interpolating missing or illiquid trading days with a zero-order hold