Hide/Show Apps

Multicriteria portfolio optimization

Tuncer Şakar, Ceren
Portfolio optimization is the problem of allocating funds between available investment options in the financial market. This thesis develops several approaches to multicriteria portfolio optimization. The use of multiple criteria is justified by demonstrating their effects on decision and objective spaces of the problem. The performance of a genetic algorithm with two and three criteria is studied; and a preference-based genetic algorithm to solve portfolio optimization with complicating constraints is developed. Furthermore, stochastic programming is used to handle multi-period problems, and several issues are studied with this approach. Efficient market hypotheses, random walk and single index models are discussed in the context of scenario generation for the Turkish Stock Market. An interactive approach to stochastic programming-based portfolio optimization is also developed to guide the decision maker toward preferred solutions. The approaches are experimented with and demonstrated using stocks from the Turkish Stock Market.