Time-Varying Regression Model with Unknown Time-Volatility for Nonstationary Signal Analysis

M. Markov (USA), O. Krasotkina, V. Mottl (Russia), and I. Muchnik (USA)

Keywords

Timevarying regression, leaveoneout principle, Kalman filter, style analysis of investment portfolios.

Abstract

The problem of estimating time-varying regression is studied via a mathematical formulation of a class of nonsta tionary signal analysis problems. This problem inevitably concerns the necessity to choose the appropriate level of model volatility – ranging from the full stationarity of in stant regression models to their absolute independence of each other. A modification of the Kalman-Bucy filter and smoother is considered to allow for easy computing of the leave-one-out error of the signal model as a quality indicator of each specific volatility level. On the basis of the proposed technique we develop a new approach to the problem of detecting the hidden dynamics of an investment portfolio in respect to certain market or economic factors.

Important Links:



Go Back