A COMPARISON OF MARKOTIWZ AND DEEP REINFORCEMENT LEARNING METHODS IN THE OPTIMIZING OF REAL ESTATE INVESTMENT FUND PORTFOLIOS
DOI:
https://doi.org/10.56238/rcsv16n3-005Keywords:
Portfolio Optimization, Modern Portfolio Theory, Real Estate Investment Trusts, Artificial Intelligence, Deep Reinforcement LearningAbstract
Portfolio optimization aims to balance the risks and returns of different assets by creating an efficient portfolio that is essential for investment management. The Brazilian real estate fund market (FIIs) is a complex task due to the heterogeneity of liquidity and the influence of macroeconomic factors. This work compares and develops portfolio optimization based on two approaches: the classic Markowitz model and Deep Reinforcement Learning (DRL). Historical data of FIIs listed in B3 between 2020 and 2024 are included, with minimum liquidity of R$ 1 million/day. The results demonstrate the viability of the solution of Learned by Reforço Profundo, embora exatas solutions ainda presentem better results in the short term.
Downloads
References
BODIE, Z.; KANE, A.; MARCUS, A. J. Investments. McGraw-Hill Education, p. 242–322, 2014.
DENG, Y. et al. Deep direct reinforcement learning for financial signal representation and trading. IEEE Transactions on Neural Networks and Learning Systems, v. 28, n. 3, p. 653–664, 2017.
GU, S.; KELLY, B.; XIU, D. Empirical asset pricing via machine learning. The Review of Financial Studies, v. 33, n. 5, p. 2223–2273, 2020.
HILL, A. Stable-Baselines3: Reliable Reinforcement Learning Implementations. Disponível em: https://stable-baselines3.readthedocs.io/en/master. Acesso em: 01 de outubro de 2025.
MARKOWITZ, H. Portfolio selection. The Journal of Finance BV, p. 77–91, 1952.
MARTIN, R. PyPortfolioOpt: Portfolio Optimization in Python. Disponível em: https://pyportfolioopt.readthedocs.io/en/latest/UserGuide.html. Acesso em: 01 de outubro de 2025.
SHARPE, W. F. Mutual fund performance. Journal of Business, v. 39, n. 1, p. 119–138, 1966.
SUTTON, R. S.; BARTO, A. G. Reinforcement Learning: An Introduction. Cambridge, MA: MIT Press, 2018.
TOWERS, M. et al. Gymnasium: A Standard Interface for Reinforcement Learning Environments. 2025. Disponível em: https://arxiv.org/abs/2407.17032. Acesso em: 02 de março de 2026.
ZHANG, G.; PATUWO, B. E.; HU, M. Y. Forecasting with artificial neural networks: The state of the art. International Journal of Forecasting, v. 14, p. 35–62, 03 1998.
Downloads
Published
Issue
Section
License
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
