On the possibilistic mean value and variance of multiplication of fuzzy numbers (Q843139)

In 2001 Carlson et al. introduced the concepts of ``possibilistic'' mean value, variance, and covariance to deal with fuzzy numbers. Those notions were next applied by other authors to solve Markowitz' mean-variance portfolio selection problem with fuzzy rates of return. In this paper the authors approach the portfolio selection problem with fuzzy investment horizon, pointing out in the introduction that most investors after having opened their positions on a capital market do not know for sure when they will close their positions. Just because of that, the resulting rates of return must be fuzzy because they depend on the length of an investment horizon which by the assumption made is a fuzzy number. A few approaches have already been proposed by other authors to perform fuzzy arithmetical operations. In this paper, the authors offer new definitions of possibilistic mean value, variance, and covariance involving double integrals which were originally mentioned by them during an IEEE international conference on fuzzy systems held in 2009. Next they reformulate the mean-variance portfolio selection problem with a fuzzy exit time and illustrate it with the example studied in 2006 by Martellini et al. in Management Science 52 where the exit time was a random variable. The example involved 5 securities from Shanghai Stock Exchange in the period of July 2000 through July 2003. The comparison of those two approaches (fuzzy exit time versus a random one) demonstrates that the minimum variance portfolio is less risky in the fuzzy model than in the probabilistic model studied by Martellini.