• Phone +44-020-7040-4575
  • Fax + 44-020-7040-8580

Dependencies across different assets are traditionally studied by analyzing the variance-covariance matrix. On a high-frequency scale, financial time series are not homogeneous, therefore standard correlation measures, like Pearson, can not be directly applied to the raw data.

To deal with this problem the time series have to be either homogenized through interpolation or methods that can handle raw non-synchronous time series need to be employed. In the following two papers we have compared two traditional methods that use interpolation with the Fourier method (Malliavan and Mancino (2002)) that can be applied directly to the actual time series and has the advantage of being model independent.

The three methods have been tested on simulated data and actual trades time series and the corresponding correlation matrices analyzed using techniques from random networks and random matrix theory. The analysis shows that the Fourier method is better than the alternatives in terms of generating smooth, robust estimates.