Nonlinear Filter Estimation of Volatility

Abstract

A discrete time nonlinear filter is used to estimate the volatility in a financial model. New filters are derived for sums of unobserved quantities and the EM algorithm applied to determine the parameters of the model.

Keywords:

• EM algorithm
• Nonlinear filters
• Reference probability approach
• Time series
• Volatility

Mathematics Subject Classification:

• Primary 60-XX, 62-XX, 91-XX
• Secondary 62M10, 62P20, 91B28, 91B84

R. E. wishes to thank the ARC and SSHRC for continued support.