A new genetic algorithm for portfolio optimization with cardinality constraints / Um novo algoritmo genetico para a otimização de carteiras de investimento com restrições de cardinalidade

AUTOR(ES)
DATA DE PUBLICAÇÃO

2008

RESUMO

In this work we consider the problem of determining of the efficient frontier of a portfolio using the mean-variance model subject to a cardinality constrain and to lower bounds on the amount invested in the selected assets. As this nonlinear integer programming problem is hard to solve exactly, we use a genetic algorithm, following the lines described by Chang et al. [3], still considered as a reference in the field. However, as the feasible solutions generated by the algorithm of Chang et al. are not uniformly distributed over the solution set, we introduce a new scheme for defining the chromosomes, based on the discretization of the feasible region, so that the amount invested always sum up to one for every solution obtained by the algorithm. This new approach allows us to define very efficient selection, crossover and mutation procedures. The numerical results obtained so far show that the new method is more robust than the one proposed by Chang et al

ASSUNTO(S)

restrições de cardinalidade portfolio optimization algoritmos geneticos genetic algorithms cardinality constraints otimização de carteiras de investimento

ACESSO AO ARTIGO

Documentos Relacionados