Distributionally Robust Portfolio Optimization with Rachev ratio using KL divergence

Document Type : Original Article

Authors
1 PhD Candidate of Industrial Management, Department of Management, Dehaghan Branch, Islamic Azad University, Dehaghan, Iran.
2 Associate Professor. Department of Management ,Dehaghan Branch, Islamic Azad University, Dehaghan, Iran.
3 Assistant Professor. Department of Industrial Engineering,Dehaghan Branch, Islamic Azad University, Dehaghan, Iran
10.22034/jik.2026.24006
Abstract
The return distribution of a stock portfolio is not constant in different periods of time, which is affected by the dynamics of financial markets and provides the basis for the instability of the stock portfolio. Distributionally Robust Portfolio Optimization (DRO) takes into account the uncertainty of the stock portfolio due to changes in the distribution of portfolio returns. In the current research, the objective function of the stock portfolio model is to maximize the Rachev ratio, which is one of the reward-risk ratios, and its calculation depends on the distribution of the stock portfolio returns. The research strategy to robust the return distribution parameter is to consider all the returns that are located in a neighborhood of the empirical distribution of the portfolio, which was used to determine such distributions using the Kl divergence. A sample portfolio of the research consists of 8 indices or industries from the Tehran Stock Exchange in the period from 1390 to 1400 and on a weekly time horizon. The test data has been divided into 5 periods, and to evaluate the DRO portfolio compared to the portfolio without this feature, the result of dividing the average of Rachev ratios in the 5 mentioned periods by their standard deviation has been used. The results show that the DRO portfolio improves this ratio by 0.40 and in addition, the minimum ratio of Rachev in 5 periods in the DRO portfolio is higher than the basket without this property.
Keywords

  • Bardakci, I. E., & Lagoa, C. M. (2019). Distributionally robust portfolio optimization. In 2019 IEEE 58th Conference on Decision and Control (CDC)(pp. 1526-1531). IEEE.
  • Du, N., Liu, Y., & Liu, Y. (2020). A new data-driven distributionally robust portfolio optimization method based on wasserstein ambiguity set. IEEE Access9, 3174-3194.
  • Fan, Z., Ji, R., & Lejeune, M. (2023). Distributionally Robust Portfolio Optimization under Marginal and Copula Ambiguity. Available at SSRN 4300358.‏
  • Hosseini-Nodeh, Z., Khanjani-Shiraz, R., & Pardalos, P. M. (2022). Distributionally Robust Portfolio Optimization with Second-Order Stochastic Dominance Based on Wasserstein Metric. Information Sciences.
  • Hu, Z., & Hong, L. J. (2013). Kullback-Leibler divergence constrained distributionally robust optimization. Available at Optimization Online, 1695-1724.
  • Ji, R., Lejeune, M. A., & Fan, Z. (2022). Distributionally robust portfolio optimization with linearized STARR performance measure. Quantitative Finance, 1-15.
  • Kobayashi, K., Takano, Y., & Nakata, K. (2021). Cardinality-constrained Distributionally Robust Portfolio Optimization. arXiv preprint arXiv:2112.12454.
  • Kobayashi, K., Takano, Y., & Nakata, K. (2023). Cardinality-constrained distributionally robust portfolio optimization. European Journal of Operational Research309(3), 1173-1182.‏
  • Liu, W., Yang, L., & Yu, B. (2021). KDE distributionally robust portfolio optimization with higher moment coherent risk. Annals of Operations Research307(1), 363-397.
  • Taghizadeh Yazdi, M., Fallahpour, S., Ahmadi Moghaddam, M. (2017). Portfolio selection by means of Meta-goal programming and extended lexicograph goal programming approaches. Financial Research Journal, 18(4), 591-612.
  • Tehrani, R., Fallah Tafti, S., & Asefi, S. (2018). Portfolio optimization using krill herd metaheuristic algorithm considering different measures of risk in Tehran stock exchange. Financial research journal, 20(4), 409-426.
  • Zhang, X. (2022). Distributional Robust Portfolio Construction based on Investor Aversion. arXiv preprint arXiv:2203.13999.