An incremental bundle method for portfolio selection problem under second-order stochastic dominance (Q2200800)

The authors consider the nonsmooth nonlinear model of the stochastic optimization problem with SSD (the second-order stochastic dominance) constraints, and no other assumptions are required. The wide application of SSD is due to its consistency with risk-averse preferences. In portfolio selection, SSD is of interest because that, for any decision-maker with non-satiable preferences and risk-averse, a portfolio which dominates a benchmark portfolio, is preferred to the benchmark. The authors propose an inexact proximal bundle method based on the incremental technique and improvement function with inexact oracle for portfolio selection problem under the SSD. The constrained nonsmooth problems are rather complex. One of the most common approaches is to solve an equivalent unconstrained problem with an exact penalty objective function. This approach, however, presents some drawbacks, which are typical whenever a penalty function is employed. To handle the constraints, the improvement function is introduced in the paper. As a result, the ``incremental'' linearizations of the forms with inexact oracle are defined. The papers is focused on ``incremental concept'' and inexact oracle for the stochastic problem, which gives a good compromise between accuracy and solvability. Some numerical experiments on portfolio optimization problems are reported.