Robust portfolio optimization with fuzzy TODIM, genetic algorithm and multi-criteria constraints

Abstract

This paper adopts the multi-criterion decision-making model of fuzzy-TODIM and geneticalgorithm (GA) for optimal portfolio allocation. We applied Markowitz’s portfolio parame-ters as inputs for the fuzzy TODIM model to rank stocks that are constituents of each indexfrom three different markets. Portfolios are then generated dynamically using three weightingtechniques and subject to multi-objective criteria and additional constraints. The results indi-cate a significant variation in performance metrics between the model-generated portfoliosand the market indices. Replication of the procedure produces a similar outcome. Moreover,the out-of-sample tests conducted over 3 years validate the results’ robustness, indicating thatfuzzy TODIM, combined with GA, can achieve superior performance in dynamic portfolioallocation.