Optimization of yield and risk of portfolio investment with use of multidimensional adaptive ARCH-models

Abstract

Building of portfolio of securities is a key decision-making task in the investment activity in stock market. In the present work classical approach of H. Markowitz to solve this task is examined, its disadvantages are displayed. Multidimensional model of autoregressive conditional heteroskedasticity is suggested, which allows to receive predictive values of variances of yields of separate assets, as well as their covariations. The possibility of extension of suggested model with adaptive mechanism is shown.