Real-Time Risk-Aversion-Adaptive Portfolio Selection by Learning a Hypernetworks
Main Article Content
Abstract
Serving a Markowitz efficient frontier at a varying risk-aversion level requires one quadratic program per query. We study a learned alternative: a hypernetwork that maps a normalized preference vector, read as a risk-aversion level, to a long-only portfolio in one forward pass. A softmax output layer enforces the budget and no-short-selling constraints, and training minimizes a Chebyshev scalarization without any pre-solved frontier. The architecture, the scalarizations and the guarantees for the scalarized problem are those of controllable Pareto front learning [1, 2]. Our contribution is an exact-oracle evaluation: for the long-only mean–variance problem the critical-line algorithm yields the exact Chebyshev solution at every preference, so the error of the
learned map can be measured exactly, compared with the strongest alternatives and located. On a ten-industry synthetic instance and a fifteen-asset VN30 instance the map attains a mean Euclidean weight error of 0.0083 and 0.0088 over ten seeds and answers a query in 44 microseconds on one CPU core. From five to two hundred assets the error grows from 0.0041 to 0.0734; its largest values occur where the highest expected returns are nearly tied, it concentrates at changes of the active set (median per-seed ratio 2.45 on VN30), and supervised training on exact targets reduces it four- to six-fold. The exact solver computes the entire frontier in 0.22 to 3.35 ms, so on this problem the learned map offers no computational advantage.
Keywords
Chebyshev scalarization, Efficient frontier, Hypernetwork, Mean–variance portfolio selection, Pareto front learning.
Article Details
This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.
References
[2] H. Markowitz, Portfolio selection: efficient diversification of investments, New York: John Wiley & Sons, 1959.
[3] A. Salo, M. Doumpos, J. Liesiö, and C. Zopounidis, Fifty years of portfolio optimization, European Journal of Operational Research, vol. 318, no. 1, pp. 1–18, Oct. 2024. https://doi.org/10.1016/j.ejor.2023.12.031
[4] H. Keffert, Robo-advising: optimal investment with mismeasured and unstable risk preferences, European Journal of Operational Research, vol. 315, no. 1, pp. 378–392, May 2024. https://doi.org/10.1016/j.ejor.2023.12.002
[5] B. Amos, Tutorial on amortized optimization, Foundations and Trends in Machine Learning, vol. 16, no. 5, pp. 592–732, Jun. 2023. https://doi.org/10.1561/2200000102
[6] D. Ha, A. M. Dai, and Q. V. Le, Hypernetworks, in Proc. 5th Int. Conf. Learn. Represent. (ICLR), 2017. https://openreview.net/forum?id=rkpACe1lx
[7] A. Navon, A. Shamsian, G. Chechik, and E. Fetaya, Learning the Pareto front with hypernetworks, in Proc. 9th Int. Conf. Learn. Represent. (ICLR), 2021. https://openreview.net/forum?id=NjF772F4ZZR
[8] L. P. Hoang, D. D. Le, T. A. Tuan, and T. N. Thang, Improving Pareto front learning via multi-sample hypernetworks, in Proc. AAAI Conf. Artif. Intell. vol. 37, no. 7, June 2023, pp. 7875–7883. https://doi.org/10.1609/aaai.v37i7.25953
[9] M. Ruchte and J. Grabocka, Scalable Pareto front approximation for deep multi-objective learning, in Proc. IEEE Int. Conf. Data Min. (ICDM), Dec. 2021, pp. 1306–1311. https://doi.org/10.1109/ICDM51629.2021.00162
[10] T. A. Tuan, L. P. Hoang, D. D. Le, and T. N. Thang, A framework for controllable Pareto front learning with completed scalarization functions and its applications, Neural Networks, vol. 169, pp. 257–273, Jan. 2024. https://doi.org/10.1016/j.neunet.2023.10.029
[11] T. A. Tuan, N. V. Dung, and T. N. Thang, A hyper-transformer model for controllable Pareto front learning with split feasibility constraints, Neural Networks, vol. 179, Art. 106571, Nov. 2024. https://doi.org/10.1016/j.neunet.2024.106571
[12] Z. Feinstein and B. Rudloff, Deep learning the efficient frontier of convex vector optimization problems, Journal of Global Optimization, vol. 90, no. 2, pp. 429–458, Oct. 2024. https://doi.org/10.1007/s10898-024-01408-x
[13] P. Chatigny, I. Sergienko, R. Ferguson, J. Weir, and M. Bergeron, Learning the efficient frontier, in Adv. Neural Inf. Process. Syst. vol. 36, 2023, pp. 22003–22020. https://doi.org/10.52202/075280-0965
[14] X. Warin, Deep learning for efficient frontier calculation in finance, Journal of Computational Finance, vol. 26, no. 1, Jun. 2022.
https://doi.org/10.21314/JCF.2021.017
[15] H. Markowitz, The optimization of a quadratic function subject to linear constraints, Naval Research Logistics Quarterly, vol. 3, no. 1–2, pp. 111–133, Mar. 1956. https://doi.org/10.1002/nav.3800030110
[16] A. Niedermayer and D. Niedermayer, Applying Markowitz’s critical line algorithm, in Handbook of portfolio construction: contemporary applications of Markowitz techniques, J. B. Guerard Jr. (Ed.), Boston, MA: Springer, 2010, pp. 383–400.
https://doi.org/10.1007/978-0-387-77439-8_12
[17] K. Miettinen, Nonlinear multiobjective optimization, vol. 12, New York, NY: Springer, 1998. https://doi.org/10.1007/978-1-4615-5563-6
[18] T. N. Thang and T. N. Hai, Self-adaptive algorithms for quasiconvex programming and applications to machine learning, Computational and Applied Mathematics, vol. 43, no. 4, Art. 249, Jun. 2024. https://doi.org/10.1007/s40314-024-02764-w
[19] X. Lin, X. Zhang, Z. Yang, F. Liu, Z. Wang, and Q. Zhang, Smooth Tchebycheff scalarization for multi-objective optimization, in Proc. 41st Int. Conf. Mach. Learn. (ICML), Proceedings of Machine Learning Research, vol. 235, 2024, pp. 30479–30509.
https://proceedings.mlr.press/v235/lin24y.html
[20] G. Cybenko, Approximation by superpositions of a sigmoidal function, Mathematics of Control, Signals, and Systems, vol. 2, no. 4, pp. 303–314, Dec. 1989. https://doi.org/10.1007/BF02551274
[21] M. Leshno, V. Y. Lin, A. Pinkus, and S. Schocken, Multilayer feedforward networks with a nonpolynomial activation function can approximate any function, Neural Networks, vol. 6, no. 6, pp. 861–867, 1993. https://doi.org/10.1016/S0893-6080(05)80131-5
[22] T. Galanti and L. Wolf, On the modularity of hypernetworks, in Adv. Neural Inf. Process. Syst. vol. 33, 2020, pp. 10409–10419.
https://proceedings.neurips.cc/paper/2020/hash/ 75c58d36157505a600e0695ed0b3a22d-Abstract.html.
[23] E. Zitzler and L. Thiele, Multiobjective evolutionary algorithms: a comparative case study and the strength Pareto approach, IEEE Transactions on Evolutionary Computation, vol. 3, no. 4, pp. 257–271, Nov. 1999. https://doi.org/10.1109/4235.797969
[24] A. F. T. Martins and R. F. Astudillo, From softmax to sparsemax: a sparse model of attention and multi-label classification, in Proc. 33rd Int. Conf. Mach. Learn.(ICML), Proceedings of Machine Learning Research, vol. 48, 2016, pp. 1614–1623. https://proceedings.mlr.press/v48/martins16.html
[25] M. J. Best and R. R. Grauer, On the sensitivity of mean-variance-efficient portfolios to changes in asset means: some analytical and computational results, The Review of Financial Studies, vol. 4, no. 2, pp. 315–342, Apr. 1991. https://doi.org/10.1093/rfs/4.2.315