New approximation for GARCH parameters estimate
arXiv:1703.03004
Abstract
This paper presents a new approach for the optimization of GARCH parameters estimation. Firstly, we propose a method for the localization of the maximum. Thereafter, using the methods of least squares, we make a local approximation for the projection of the likelihood function curve on two dimensional planes by a polynomial of order two which will be used to calculate an estimation of the maximum.
10 pages, 4 tables, 3 figures