Covariance matrix forecasting using support vector regression

Abstract Support vector regression is a promising method for time-series prediction, as it has good generalisability and an overall stable behaviour. Recent studies have shown that can describe the dynamic characteristics of financial processes make more accurate forecasts than other machine learning techniques. The first main contribution this paper to propose methodology modelling forecasting covariance matrices based on support using Cholesky decomposition. procedure applied range-based returns, which are estimated basis low high prices. Such prices most often available with closing many series contain information about volatility relationships between returns. guarantees positive definiteness forecasted flexible, be different dependence patterns. second show example exchange rates from forex market matrix calculated proposed approach benchmark conditional correlation model. advantage suggested higher during turbulent periods, i.e., when difficult matter most.