Title: Portfolio diversification strategy via ARMA-GARCH vine copula approach
Authors: Hao Ji - Northwest Agricultural and Forestry University (China) [presenting]
Abstract: Under the framework of the Black-Litterman model, the copula-opinion pooling method (COP method) does not consider heteroscedasticity and autocorrelation of assets, and is not flexible enough to describe the dependence structure compared with the vine copula in high-dimensional cases. A portfolio diversification application is extended to the ARMA-GARCH-Vine-Copula approach in order to overcome the drawbacks of COP method. First, the ARMA-GARCH model is used for each univariate returns series to remove the heteroscedasticity and autocorrelation. Then, conditional Spearman's has been chosen as the measurement to classify clusters of candidate assets. Algorithms for the numerical estimation of conditional Spearman's are also illustrated. After performing the cluster algorithms, a vine copula is used to model the dependence structure among assets returns. We choose one asset from each cluster as a candidate asset to avoid co-moving in their lower regions. Finally, a strategy is presented to construct diversified portfolios at one day forecast horizon by adding the investors information into portfolio selection procedure and minimizing the CVaR. It can be considered as an extension of the Black-Litterman model. The presented methodology is expected to be useful for the selection of a diversified portfolio of asset returns.