[S] Estimation problems with large covariance matrices

frank.haertel@ubs.com
Fri, 17 Jul 1998 11:10:32 +0200


This is a statistical issue partly dealing with S-Plus and hopefully
somebody in the S-Plus community can help me on that:

We want to estimate covariance matrices with approximately 1000-2000
variables (e.g. securities). One question now is how many observations
are needed for estimating the full covariance matrix?

Does a S-Plus function exist (or has anybody written one) on
estimating the number of observations needed, based on the number of
variables to be estimated (especially for large variable sets)?

Does anybody have experiences with estimating such large covariance
matrices? Do special problems arise (positive-definiteness, many small
eigenvalues, "noisy" results) and are there any statistical (S-Plus
based) methods on dealing with these situations?

Thank you very much for your help.

--------------------------------------------
Dr. Frank Haertel
UBS Brinson
Asset Allocation/Risk Investment Research
Gartenstr. 9/1A
4002 Basel
Switzerland

Tel.: ++41/61/288 63 87
FAX: ++41/61/288 44 24
E-Mail: frank.haertel@ubs.com
---------------------------------------------

-----------------------------------------------------------------------
This message was distributed by s-news@wubios.wustl.edu. To unsubscribe
send e-mail to s-news-request@wubios.wustl.edu with the BODY of the
message: unsubscribe s-news