Portfolio selection based on predictive joint return distribution

A predictive joint return distribution can provide more useful information than moment-based risk measures in portfolio selection. This article develops a D-vine copula-CAViaR method to estimate and predict the joint probability distribution of multiple financial returns. Furthermore, we construct a portfolio model via the generalized Omega ratio inferred from the predicted joint return distribution. The superiority of our method is illustrated through an empirical application on five international stock market indices.