Title: Copula-based multiobjective portfolio optimization
Authors: Maziar Sahamkhadam - Linnaeus University (Sweden) [presenting]
Abstract: Utilizing the multicriteria decision making (MCDM) and vine copulas, a copula-based multiobjective portfolio (MOP) optimization is developed. Using the copula-based MOP optimization model, we evaluate the impact of objective functions on portfolio performance. Furthermore, we compare the copula-based MOP with those obtained from two sophisticated predictive models, including the multivariate GARCH and the factor stochastic volatility. Applying the MOP optimization to a sample of S\&P 100 equities, an in-sample analysis of the Pareto sets reveals that the risk models generally perform better than a historical-simulation approach in generating optimal efficient sets when there are higher preferences on return and tail risk. Based on an out-of-sample analysis, the predictive models generate multicriteria portfolios that achieve statistically significant improvements concerning return and tail risk during a market downturn such as the global financial crisis (GFC) and the COVID-19 market crash. Overall, there is evidence that the copula-based multicriteria portfolios perform better than those from the other predictive models in terms of the downside risk. With a view to the portfolio attributes, the dividend yield and beta coefficient show significant negative impacts on portfolio tail risk measures.