Robust portfolio management
In financial engineering and asset management, the most classical real-world application is Markowitz’s mean-variance portfolio optimization [Mar52]. Investors and fund managers seek to simultaneously maximize expected returns and minimize risk (often modeled as the variance of returns). This framework is widely utilized by institutional pension funds (e.g., GPIF), treasury departments, and risk management teams to quantify trade-offs and build consensus on asset allocation decisions.
However, in practice, estimating the covariance matrix from limited observations or highly collinear assets often leads to numerical instability. To resolve this, it is common to introduce a strongly convex regularization term, such as an \(L_2\) norm penalty on the asset allocation weights or a turnover penalty to suppress excessive trading compared to a previous portfolio \(x^{\text{prev}}\) [Qi].
This formulation effectively creates a three-objective optimization problem over the allocation weights \(x \in \mathbb{R}^n\):
Subject to classical constraints such as the budget constraint \(\mathbf{1}^T x = 1\) and long-only constraints \(x \ge 0\).
Because the regularization term is strongly convex, the resulting scalarized objective function becomes strictly strongly convex. Even if the covariance matrix \(\Sigma\) is only positive semi-definite, the addition of the turnover penalty guarantees that the Hessian \(\nabla^2 f(x) \succeq 2\lambda I\) is strictly positive definite. According to the theorems in [MHI21], this strongly convex problem is guaranteed to be weakly simplicial.
Fitting a Bézier simplex to this problem allows practitioners to continuously map the entire robust Pareto front—namely, the continuous efficient frontier—guaranteeing unique solutions. Instead of computing disconnected discrete point clouds using weighted sums, analysts obtain a functionally continuous mapping of allocations versus risk. Utilizing sensitivity-based Newton path-following, the full Pareto front can be computed with an extremely efficient iteration complexity of \(O(p \log(1/\varepsilon))\) [BDK21]. Solvers like MOSEK, Gurobi, and OSQP are widely used to efficiently compute these strongly convex QP subproblems. Modern extensions also include CVaR (Conditional Value-at-Risk) portfolio optimization and robust multi-objective portfolio frameworks under data uncertainty.
References
El Houcine Bergou, Youssef Diouane, and Vyacheslav Kungurtsev. Complexity iteration analysis for strongly convex multi-objective optimization using a newton path-following procedure. Optimization Letters, 15:1215–1227, 2021. doi:10.1007/s11590-020-01623-x.
Henri Bonnel and Corinne Schneider. Post-pareto analysis and a new algorithm for the optimal parameter tuning of the elastic net. Journal of Optimization Theory and Applications, 183(3):993–1027, 2019. doi:10.1007/s10957-019-01592-x.
S Breedveld, P Storchi, M Keijzer, A Heemink, and B Heijmen. A novel approach to multi-criteria inverse planning for imrt. Physics in Medicine and Biology, 2007.
Olivier de Weck and Il Yong Kim. Adaptive weighted sum method for multiobjective optimization. Submitted to Structural and Multidisciplinary Optimization for Review.
Jörg Fliege, L M Graña Drummond, and Benar Fux Svaiter. Newton's method for multiobjective optimization. SIAM Journal on Optimization, 20(2):602–626, 2009. doi:10.1137/08071692X.
Naoki Hamada. 多目的最適化の解集合のトポロジーの検定法 (statistical test for the topology of solution sets in multi-objective optimization). In 日本応用数理学会2020年度年会講演予稿集 (Proceedings of the JSIAM Annual Meeting 2020). 2020.
Naoki Hamada and Keisuke Goto. Data-driven analysis of pareto set topology. In Proceedings of the Genetic and Evolutionary Computation Conference, 657–664. 2018. URL: https://doi.org/10.1145/3205455.3205613.
Naoki Hamada, Kenta Hayano, Shunsuke Ichiki, Yutaro Kabata, and Hiroshi Teramoto. Topology of pareto sets of strongly convex problems. SIAM Journal on Optimization, 30(3):2659–2686, 2020. doi:10.1137/19M1264175.
Per Christian Hansen and Dianne Prost O'Leary. The use of the l-curve in the regularization of discrete ill-posed problems. SIAM Journal on Scientific Computing, 14(6):1487–1503, 1993. doi:10.1137/0914086.
Ken Kobayashi, Naoki Hamada, Akiyoshi Sannai, Aiko Tanaka, Kenichi Bannai, and Masashi Sugiyama. Bézier simplex fitting: describing pareto fronts of simplicial problems with small samples in multi-objective optimization. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 33, 2304–2313. 2019. URL: https://doi.org/10.1609/aaai.v33i01.33012304.
Suyun Liu and Luís N Vicente. The stochastic multi-gradient algorithm for multi-objective optimization and its application to supervised machine learning. Annals of Operations Research, 339(3):1119–1148, 2024. doi:10.1007/s10479-021-04033-z.
Harry Markowitz. Portfolio selection. The Journal of Finance, 7(1):77–91, 1952. doi:10.2307/2975974.
Yuto Mizota, Naoki Hamada, and Shunsuke Ichiki. All unconstrained strongly convex problems are weakly simplicial. 2021. URL: https://arxiv.org/abs/2106.12704, arXiv:2106.12704.
Houduo Qi. Optimal portfolio selections via l1,2-norm regularization. URL: https://eprints.soton.ac.uk/451606/1/mvpl12_for_PURE.pdf.
Robert Tibshirani, Michael Saunders, Saharon Rosset, Ji Zhu, and Keith Knight. Sparsity and smoothness via the fused lasso. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 67(1):91–108, 2005.
Dongxiang Yan, He Yin, Tao Li, and Chengbin Ma. A two-stage scheme for both power allocation and ev charging coordination in a grid tied pv-battery charging station. IEEE Transactions on Industrial Informatics, 2021. doi:10.1109/TII.2021.3054417.
Ming Yuan and Yi Lin. Model selection and estimation in regression with grouped variables. Journal of the Royal Statistical Society: Series B (Statistical Methodology), 68(1):49–67, 2006.