Schur Complementary Portfolios
A bridge between hierarchical and optimisation-based portfolio construction.
Portfolio construction has long had two camps. Top-down methods build a hierarchy of assets and allocate between and within clusters — robust, intuitive, and forgiving of a noisy covariance estimate. Bottom-up methods solve a single optimisation over the whole covariance matrix — efficient when the estimate is good. Both are reasonable; they simply trust the data to different degrees.
The Schur complement connects them. Augment each cluster's covariance with the Schur complement of the others before allocating, and the hierarchical construction reproduces the global optimisation exactly. A single dial $\gamma \in [0, 1]$ slides smoothly between the two worlds — and, encouragingly, the best out-of-sample portfolios usually live somewhere in between.
One dial: $\gamma = 0$ recovers hierarchical risk parity, $\gamma = 1$ recovers the minimum-variance portfolio, and the interior interpolates between them.
For the derivation — the block-inverse identity, the $\gamma$ interpolation, and the recursion at arbitrary hierarchy depth — see the introduction or the paper.
Implementations
Three maintained libraries cover the batch, streaming, and estimation layers: skfolio (fixed universe, sklearn idiom), allocation (online, evolving universe, Fiedler seriation), and precise (online covariance and the Schur pseudo-likelihood). Install commands, repositories, and tutorials are on the implementations page.
Further reading
- Introduction — the method explained step by step on this site, with the block-inverse identity and the $\gamma$ interpolation.
- Cotton, P. (2022). “Schur Complementary Portfolios Fix Hierarchical Risk Parity.” Geek Culture (Medium) — the original, informal introduction that started it all.
- Cotton, P. (2024). arXiv:2411.05807 — the canonical paper.
- CQF slides — a compact derivation with the interpolation argument.
- Palomar, D. P. Portfolio Optimization, §12.3.4 — textbook treatment alongside HRP and the global minimum-variance portfolio.
Cite
Cotton, P. (2024). “Schur Complementary Allocation: A Unification of Hierarchical Risk Parity and Minimum Variance Portfolios.” arXiv preprint arXiv:2411.05807.
@article{cotton2024schur,
author = {Cotton, Peter},
title = {Schur Complementary Allocation: A Unification of Hierarchical
Risk Parity and Minimum Variance Portfolios},
journal = {arXiv preprint arXiv:2411.05807},
year = {2024},
url = {https://arxiv.org/abs/2411.05807}
}
Bibliography
Works on, citing, or directly extending Schur complementary allocation — plus the spatial-statistics side of the same idea. Every paper below also appears on the literature map; the dated story is on the timeline.
The core
- Cotton, P. (2022). “Schur Complementary Portfolios Fix Hierarchical Risk Parity.” Geek Culture (Medium). Where the γ-bridge first appeared, November 2022.
- Cotton, P. (2023). Schur Complementary Portfolios — CQF slides. PDF. Compact derivation with the interpolation argument.
- Cotton, P. (2024). Schur Complementary Allocation: A Unification of Hierarchical Risk Parity and Minimum Variance Portfolios. arXiv:2411.05807. The canonical write-up; the recursion at arbitrary depth.
- Cotton, P. (2026). Two Sides of Schur Damping: High-Dimensional Pseudo-Likelihoods and Portfolio Allocation. arXiv:2606.14798. The cross-field identity: the same damping, and the same closed-form reliability γ*, in allocation and in spatial pseudo-likelihoods.
- Cotton, P. (2026). Schur Damping for Perpetual Demand Lending Pools. Working paper — PDF, web note. Brings the interior γ-dial to the single-pool-vs-two-pools decision of Chitra et al.; γ* set by the reliability of the cross-pool covariance, pulled toward 0 in the undersampled DeFi regime.
Antecedents
- López de Prado, M. (2016). “Building Diversified Portfolios that Outperform Out of Sample.” Journal of Portfolio Management 42(4), 59–69. Hierarchical risk parity — the γ=0 endpoint.
- Antonov, A., Lipton, A., and López de Prado, M. (2024). Overcoming Markowitz's Instability with the Help of the Hierarchical Risk Parity (HRP): Theoretical Evidence. SSRN 4748151. Analytical noise comparison of HRP vs Markowitz.
- Palomar, D. P. Portfolio Optimization, §12.3.4 “Schur complementary portfolios”. Online edition. Textbook treatment alongside HRP and global minimum variance.
Theory and extensions
- Mograby, G. (2025). Hierarchical Minimum Variance Portfolios: A Theoretical and Algorithmic Approach. arXiv:2503.12328. Exact Schur separator recursion on hierarchical graphs — the γ=1 skeleton, made rigorous.
- Wuebben, B. J. (2026). Beyond De Prado and Cotton: Hierarchical and Iterative Methods for General Mean-Variance Portfolios. arXiv:2604.23833. HRP-μ, HRP-Σμ, CRISP: signal-aware extensions past minimum variance.
- Knežević, P., and Posedel Šimović, P. (2026). Bridging Risk Parity and Variance Optimization: A Schur Complement Approach to Recursive Asset Allocation. The Journal of FinTech 2670001, open access (PDF). Cites both the 2022 blog post and the 2024 paper.
- Alonso, N. I. (2025). Return-Adjusted Hierarchical Risk Parity and Schur Portfolios: A Comprehensive Theoretical and Empirical Study. SSRN 5370624.
- Alonso, N. I. (2026). Reinforcement Learning Portfolio Optimization (RLPO): From Markowitz to Risk-Sensitive Control. SSRN 6447220.
The spatial-statistics side
- Vecchia, A. V. (1988). “Estimation and Model Identification for Continuous Spatial Processes.” JRSS-B 50(2). The neighbour-conditioned factorization; the conditionals are Schur complements.
- Katzfuss, M., and Guinness, J. (2021). “A General Framework for Vecchia Approximations of Gaussian Processes.” Statistical Science 36(1).
- Chakraborty, A., and Katzfuss, M. (2025). Learning Non-Gaussian Spatial Distributions via Bayesian Transport Maps with Parametric Shrinkage (ShrinkTM). arXiv:2409.19208. Damping arrives in spatial statistics: conditionals shrunk toward a parametric base.
Applications and empirical studies
- Chitra, T., Diamandis, T., Sheng, N., Sterle, L., and Yusubov, K. (2025). Perpetual Demand Lending Pools. arXiv:2502.06028. Appendix B decides single vs multiple DeFi lending pools via the undamped Schur complement of the pool covariance — the exact conditional covariance, no γ. The γ=0 vs γ=1 architecture question, posed for billions of dollars of pooled assets — see the note on damping in PDLPs for what the interior dial adds.
- Salas-Molina, F., Pla-Santamaria, D., Garcia-Bernabeu, A., and Reig-Mullor, J. (2025). Estimation Windows in Hierarchical Risk Parity Methods for Portfolio Selection. LNCS Decision Sciences. Empirical HRP study; best out-of-sample Sharpe near five years of daily data.
- Pergher, K. G. R., Soldera, J., and Scharcanski, J. (2026). An Orthogonal Hierarchical Risk Parity Allocation Method for Improved Portfolio Out-of-Sample Performance. IEEE Access.
- Bergmeier, J. (2026). The Risk Parity Zoo: Which Risk Contributions Should Multi-Asset Investors Equalize?. SSRN 6536678. Out-of-sample comparison of risk-based allocation models.
- Nicolini, C., Manzi, M., and Delatte, H. (2025). skfolio: Portfolio Optimization in Python. arXiv:2507.04176. The library shipping the SchurComplementary optimizer.
Citation list maintained from Google Scholar forward citations. Working on a related paper or implementation? Open an issue on the schur repo and we'll add it.