عنوان مقاله [English]
Asset allocation optimization under nonsymmetrically distributed and dependent returns is required for real world applications.
So far, Monte Carlo based scenario optimization methods have been most successful for such problems, but may be computationally expensive. Many methods that do not use simulated data exist when the random returns are normally distributed. A new computationally efficient method has been proposed namely, the separable probabilistic robust method that is not scenario based and directly maximizes the overall return for a given probability of non-exceedance(risk), namely the value-at-risk (VaR). A byproduct of this method is the conditional value-at-risk (CVaR). In addition, a new asset allocation method that uses Monte Carlo simulations is also proposed, which retains the original size of the problem, unlike many current scenario based methods. Copula is used to model joint distributions of dependent and nonsymmetrical returns. The two new methods are tested with real world stock prices and the results are comparable to the best solutions available, in terms of the objective function value, namely, the VaR.