×

Solving dynamic stochastic control problems in finance using tabu search with variable scaling. (English) Zbl 0877.90008

Osman, Ibrahim H. (ed.) et al., Meta-heuristics: theory and applications. International conference (MIC), Breckenridge, CO, USA, 22–26 July 1995. Dordrecht: Kluwer Academic Publishers. 429-448 (1996).
Summary: Numerous multistage planning problems in finance involve nonlinear and nonconvex decision controls. One of the simplest is the fixed-mix investment strategy. At each stage during the planning horizon, an investor rebalances her/his portfolio in order to achieve a target mix of asset proportions. The decision variables represent the target percentages for the asset categories. We show that a combination of tabu search and variable scaling generates global optimal solutions for real world test cases, despite the presence of nonconvexities. Computational results demonstrate that the approach can be applied in a practical fashion to investment problems with over 20 stages (20 years), 100 scenarios, and 8 asset categories. The method readily extends to more complex investment strategies with varying forms of nonconvexities.
For the entire collection see [Zbl 0869.00056].

MSC:

91B28 Finance etc. (MSC2000)
93E20 Optimal stochastic control
68T05 Learning and adaptive systems in artificial intelligence
PDFBibTeX XMLCite