Index Tracking : Genetic Algorithms for Investment Portfolio Selection
J Shapcott · 2002
This project was concerned with passive portfolio selection using genetic algorithms and quadratic programming techniques. Searching a large universal set of shares for a subset that performs well is intractable, so a stochastic search method must be used. The genetic algorithm generates the subsets, and quadratic programming is used to find both their performance and the proportion of the available capital that should be invested in each member company. Separate subpopulations are maintained on different processors of a Meiko Computing Surface, with occasional migration of genomes. This strategy allows several differing threads of the search to be pursued within the separate subpopulations, and migration encourages convergence to the global optimum, instead of local optima. 1 1