Prediction of financial risks using a modified grid method sliding separation precipitation shear mixtures of normal laws

Abstract

Describes a method for predicting financial risks using parametric models from the class of precipitation - shear mixtures of normal laws. The proposed method takes as its basis the results of the modified two - step grid method for estimating the parameters of the generalized hyperbolic distributions in the sliding mode. Discusses in detail the practical application of the method, the speed of his work. Describes how to set parameters (training model). On real data illustrates the prediction accuracy when using different metrics depending on the setting method, including long - term forecasts are considered in more than one step ahead.