{"paper_id":"2503.12328","title":"Hierarchical Minimum Variance Portfolios: A Theoretical and Algorithmic Approach","authors":["Gamal Mograby"],"paper_date":"2025-03-16","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2503.12328v1","page_url":"https://thequant.space/flowcharts/hierarchical-minimum-variance-portfolios-a-theoretical-and/","scores":{"empirical_rigor":2,"math_complexity":8.5,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a hierarchical minimum variance portfolio (HMVP) approach that models the covariance matrix as an adjacency-like matrix of a hierarchical graph and uses the Schur complement to recursively reduce the optimization problem, enabling efficient computation of optimal weights.","quote":"We introduce a novel approach to portfolio optimization that leverages hierarchical graph structures and the Schur complement method to systematically reduce computational complexity while preserving full covariance information. Moreover, we translate our results into a recursive algorithm that constructs optimal portfolio allocations.","summary":"We introduce a novel approach to portfolio optimization that leverages hierarchical graph structures and the Schur complement method to systematically reduce computational complexity while preserving full covariance information. Inspired by Lopez de Prados hierarchical risk parity and Cottons Schur…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The covariance matrix is invertible.","type":"model","stated":"explicit","quote":"We seek the optimal weight vector w that minimizes portfolio variance while ensuring that the total portfolio weight sums to one, i.e. min w n wtΣw n X i=1 wi = 1 o Under the assumption that the covariance matrix is invertible, the minimum variance portfolio is computed using: (2.5) w∗= Σ−11 1tΣ−11.","support":"full"},{"assumption":"The covariance matrix is positive definite (so it is invertible).","type":"statistical","stated":"explicit","quote":"Note that the covariance matrix Σ is positive semi-definite, meaning it is invertible if and only if it is positive definite.","support":"full"},{"assumption":"Assets are selected such that their covariance matrix corresponds to the adjacency matrix of a hierarchical graph.","type":"model","stated":"explicit","quote":"However, in our approach, we construct the portfolio by selecting assets such that their covariance matrix describes the adjacency matrix of a hierarchical graph.","support":"full"}]},"datasets":{"status":"none_found","items":[],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"none"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"none","findings":[{"finding":"The hierarchical reduction preserves full covariance information while reducing computational complexity.","kind":"theoretical","quote":"We emphasize that this approach reduces computational complexity while preserving the complete co-variance information.","support":"full"},{"finding":"Optimal portfolio weights can be computed efficiently by inverting only a series of 3 × 3 matrices, independent of portfolio size.","kind":"theoretical","quote":"Instead, we show that the optimal portfolio weights can be computed efficiently by inverting only a series of 3 × 3 matrices, a process that remains computationally feasible independent of portfolio size.","support":"full"},{"finding":"Theorem 3.7 shows that solving the weight optimization at level ℓ can be recursively reduced to an equivalent problem at level ℓ‑1.","kind":"theoretical","quote":"Our result, Theorem 3.7, states that solving the optimization problem for portfolio weights, i.e., finding wℓ= Σ−1 ℓ1, can be recursively reduced to an equivalent optimization problem on the lower-level graph Gℓ−1, expressed as wℓ−1 = Σ−1 ℓ−1γℓ−1.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":11,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":2,"extracted_at":"2026-11-14T19:51:25Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}