Stock portfolio with maximum predictability of Sharpe ratio based on hidden Markov model

Nabiyan, Ali; Baktash, Forozan; Davoodi, Sayyed Mohammad Reza

doi:10.22075/ijnaa.2024.33369.4967

Stock portfolio with maximum predictability of Sharpe ratio based on hidden Markov model

Document Type : Research Paper

Authors

Ali Nabiyan

Forozan Baktash

Sayyed Mohammad Reza Davoodi

Department of Economics, Dehagan Branch, Islamic Azad University, Dehaghan, Iran

10.22075/ijnaa.2024.33369.4967

Abstract

The Sharpe ratio is one of the performance evaluation criteria of the stock portfolio, which shows the return per unit of risk. This ratio is particularly important for risk-averse investors. In the current research, the hidden Markov model approach introduces a new stock portfolio model called the maximum predictability portfolio of the Sharpe ratio. The hidden Markov model used for each hidden state has a mixture of normals distribution output, which is used to calculate the return and standard deviation to calculate the Sharpe ratio in the investment horizon. The research portfolio calculates the weights of the portfolio in such a way that the Sharpe ratio is maximized in the horizon of the portfolio. The optimal research portfolio was optimized using the historical data of 10 indices from the Tehran Stock Exchange between 2018 and 2018 in a four-member mode space. The evaluation of the performance of the optimal portfolio in the Sharpe ratio criterion shows that the research model has a better performance than the mean-variance model and the equal-weight model.

Keywords

Sharpe ratio

Markov chain

hidden Markov model

mixed normal distribution

[1] M.W. Burkett, W.T. Scherer, and A. Todd, Portfolio design and management through state-based analytics: A probabilistic approach, Cogent Econ. Finance 8 (2020), no. 1, 1–17.

[2] P.G. Coulombe and M. Goebel, Maximally machine-learnable portfolios, arXiv preprint arXiv:2306.05568 (2023), Available at SSRN: https://ssrn.com/abstract=4428178 or http://dx.doi.org/10.2139/ssrn.4428178.

[3] R.D. Harris, J. Shen, and F. Yilmaz, Maximally predictable currency portfolios, J. Int. Money Finance 128 (2022), 102702.

[4] J. Jiang, L. Wu, H. Zhao, H. Zhu, and W. Zhang, Forecasting movements of stock time series based on hidden state guided deep learning approach, Inf. Process. Manag. 60 (2023), no. 3, 103328.

[5] W. Liu, L. Yang, and B. Yu, KDE distributionally robust portfolio optimization with higher moment coherent risk, Ann. Oper. Res. 307 (2021), no. 1, 363–397.

[6] J. Lu, Y. Hu, X. Li, Q. Zhu, H. Wu, and S. Yang, A new stock forecasting model by hidden quantum Markov models, Int. Conf. Artific. Intell. Secur., 2022, pp. 438–447.

[7] R. Majumder, Q. Ji, and N. K. Neerchal, Optimal stock portfolio selection with a multivariate hidden Markov model, Sankhya B 85 (2023), no. 1, 177–198.

[8] N. Petkovic, M. Bozinovic, and S. Stojanovic, Portfolio optimization by applying Markov chains, Anal. Ekonom. Fakul. Subot. 40 (2018), no. 2, 21–32.

[9] E.W. Park, Principal component analysis and hidden Markov model for forecasting stock returns, arXiv preprint arXiv: 2307.00459 (2023).

[10] P. Pasricha, D. Selvamuthu, G. D’Amico, and R. Manca, Portfolio optimization of credit risky bonds: A semi-Markov process approach, Financ Innov 25 (2020), no. 6, 1–14.

[11] R. Ruiz-Cruz, C. Sedano, and O. Flores, Genetic optimization of a trading algorithm based on pattern recognition, IEEE Latin Amer. Conf. Comput. Intell. (LA-CCI), 2019, pp. 1–6.

[12] H. Ryou, H.H. Bae, H.S. Lee, and K.J. Oh, Momentum investment strategy using a hidden Markov model, Sustainability 12 (2020), no. 17, 7031.

[13] Z. Su and B. Yi, Research on HMM-Based Efficient Stock Price Prediction, Mobile Information Systems, 2022.

[14] B.Q. Ta, V.T. Huynh, N.Q. Nguyen, P.N. Nguyen, and B.H. Ho, Maximal predictability portfolio optimization model and applications to Vietnam stock market, Int. Conf. Thailand Econ. Soc., 2022, pp. 559–578.