Optimization of portfolio management based on vector autoregression models and multivariate volatility models

Theoretical part of this article examines the impact of information on the stochastic model of generating returns of assets (vector autoregressive model) on the optimal structure of assets allocation of the investment portfolio. Article includes theoretical basis for construction and characteristics of the optimal portfolios. The results of simulation showed that the characteristics of the studied portfolios under certain conditions may significantly exceed the performance of classic mean-variance portfolios. The practical part examines the characteristics of optimal portfolios whose asset returns are predicted by the VAR models and the covariance matrixes of the assets using multivariate models of volatility. The results of practical studies have shown that the model volatility significantly affect the characteristics of optimal portfolios, and also confirmed the need and importance of investigating the errors characteristics of forecasts of portfolio returns.