Синтез адаптивных моделей GARCH для прогнозирования максимальных условных дисперсий невязок соотношений координат процесса с разнотемповой дискретизацией

Abstract

The concept of maximal conditional sample variance of ratio discrepancy is defined and the design method of GARCH model for variance forecast is proposed. Process dynamics in stochastic environment is described by ARMA model with multirate sampling. Adaptive tuning of GARCH model’s coefficients is based on recurrent least squares method.