Seleção e otimização de portfólios de investimentos: uma abordagem híbrida com DEA e MARKOWITZ

Abstract

This paper proposes an integrated strategy for investment portfolio construction by combining Data Envelopment Analysis (DEA) and Markowitz's Modern Portfolio Theory, aiming to overcome the limitations of traditional asset selection methods, especially in emerging markets. The methodology is divided into two main stages: (i) asset pre-selection through the DEA-BCC model, which evaluates relative efficiency based on fundamental indicators (EPS, ROE, P/E) and risk (Beta); and (ii) optimization via Markowitz's model, using Monte Carlo simulation and sequential quadratic programming (SLSQP) to define the efficient frontier. The results, applied to 84 assets from the Ibovespa, demonstrated the selection of 19 decision-making units (DMUs) with efficiency above 90%, resulting in a portfolio with an average beta of 0.715 – indicating lower systematic risk than the market. It is concluded that the proposed hybrid approach reduces subjective biases, improves diversification, and is operationally feasible, offering a robust framework for asset managers.