University of Cambridge > Talks.cam > Signal Processing and Communications Lab Seminars > Sequential Monte Carlo methods for estimating large scale volatility matrices - CANCELLED (to be rearranged in Lent term)

Sequential Monte Carlo methods for estimating large scale volatility matrices - CANCELLED (to be rearranged in Lent term)

Add to your list(s) Download to your calendar using vCal

If you have a question about this talk, please contact Rachel Fogg.

During the recent credit crunch and the aftermath of the financial crisis it has become even more evident than before, that there is a need for accurate estimation of financial uncertainty. This is usually translated into estimating the volatility or variance of associated financial instruments, such as asset returns. In particular, there is increasing interest in large data sets with the objective of identifying similarities between asset returns, pinpointing highly volatile assets and thus identifying and measuring financial risk.

In this talk, I will define a suitable stochastic volatility model based on Wishart autoregressive processes. Such processes are natural models for estimating covariance matrices and their theoretical background has only been recently developed. We propose a sequential Monte Carlo algorithm, based on an auxiliary particle filter. We discuss the problem of unknown parameter estimation in the particle filter, and we provide modifications to existing procedures to deal with the large number of such parameters. We address the problem of model comparison in this particular application by considering several alternative models. Our findings suggest that particle filters are very useful in this context, overcoming most of the shortcomings of alternative models.

This talk is part of the Signal Processing and Communications Lab Seminars series.

Tell a friend about this talk:

This talk is included in these lists:

Note that ex-directory lists are not shown.

 

© 2006-2017 Talks.cam, University of Cambridge. Contact Us | Help and Documentation | Privacy and Publicity