{"paper_id":"2512.23021","title":"Squeezed Covariance Matrix Estimation: Analytic Eigenvalue Control","authors":["Layla Abu Khalaf","William Smyth"],"paper_date":"2025-12-28","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2512.23021v1","page_url":"https://thequant.space/flowcharts/squeezed-covariance-matrix-estimation-analytic-eigenvalue-c/","scores":{"empirical_rigor":7,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a squeezed covariance estimation framework called atomic‑IQ, featuring a squeezing identity and an analytic eigenvalue floor for PSD guarantees and conditioning control.","quote":"We introduce a squeezing identity that represents IQ estimators as a convex-like combination of structured channel matrices, and propose an atomic-IQ parameterization in which each channel-class matrix is built from PSD atoms with a single class-level normalization.","summary":"We revisit Gerber’s Informational Quality (IQ) framework, a data-driven approach for constructing correlation matrices from co-movement evidence, and address two obstacles that limit its use in portfolio optimization: guaranteeing positive semidefinite ness (PSD) and controlling spectral…","strategy":"Atomic‑IQ (AIQ1)"},"assumptions":{"status":"available","items":[{"assumption":"Portfolio construction is a long‑only tangency (maximum‑Sharpe) problem.","type":"execution","stated":"explicit","quote":"... the long‑only tangency (maximum‑Sharpe) portfolio."},{"assumption":"Decisions are made monthly with a rolling look‑back window of 20 months for returns and covariance.","type":"data","stated":"explicit","quote":"At each monthly decision date d we estimate expected returns ˆµd and the covariance matrix ˆΣd from a rolling lookback window of τ=20 months, with ˆµd computed"},{"assumption":"Transaction costs are a flat 10 basis‑point charge on dollar turnover at each rebalance.","type":"execution","stated":"explicit","quote":"Transaction costs of 10 bps are charged on dollar turnover at each rebalance."},{"assumption":"The estimator remains PSD whenever the collective squeeze lies in the feasible region defined by the canonical squeezing identity.","type":"model","stated":"explicit","quote":"Channel‑class matrices are assembled from positive semi‑definite atoms and normalized at the class level, so the estimator remains PSD whenever the collective squeeze lies in the feasible region."},{"assumption":"Eigenvalue control is achieved via an analytic eigenfloor that can target a minimum eigenvalue or a desired condition number.","type":"statistical","stated":"explicit","quote":"We develop an analytic eigenfloor that targets either a minimum eigenvalue or a desired condition number and, when necessary, repairs PSD violations in closed form while remaining compatible with the squeezing identity."}]},"datasets":{"status":"available","items":[{"name":"Diversified multi‑asset universe (N=10)","provider":null,"kind":"market-prices","access":"unknown","period":"1988–2024","universe":"10‑asset portfolio","quote":"Using a diversified multi‑asset universe and a tangency (maximum‑Sharpe) objective"}],"known":[],"sample":{"start":"2000","end":"2024","market":null,"asset_class":"equities","frequency":"monthly"},"data_resolution":"weekly-monthly"},"transaction_costs":{"status":"available","modeled":"yes","headline_net_of_costs":true,"detail":"10 bps charged on dollar turnover at each monthly rebalance.","quote":"Transaction costs of 10 bps are charged on dollar turnover at each rebalance.","source":"front_matter","rule":"R3","candidates":{"front_matter":"yes","evidence_card":"yes","evidence_card_quoted":true},"card_detail":"10 bps charged on dollar turnover at each monthly rebalance.","card_quote":"Transaction costs of 10 bps are charged on dollar turnover at each rebalance.","meaning":"costs are part of the reported headline results"},"results":{"status":"available","evaluation_type":"trading-pnl","findings":[{"finding":"AIQ1 achieves the highest out‑of‑sample Sharpe ratio of 0.56, with AIQ2 close behind at 0.54.","kind":"performance","quote":"AIQ1 delivers the highest out‑of‑sample Sharpe ratio (0.56), with AIQ2 close behind (0.54)."},{"finding":"AIQ1 and AIQ2 exhibit lower realized volatility (7.70% and 7.65%) and moderate drawdowns (~‑28%) compared to higher‑volatility shrinkage alternatives.","kind":"performance","quote":"This risk‑adjusted performance is achieved with comparatively low volatility (7.70% for AIQ1 and 7.65% for AIQ2), and with moderate drawdowns (approximately ‑28%) relative to the higher‑volatility shrinkage alternatives."},{"finding":"AIQ1 attains the best overall aggregate score (34) across multiple risk‑return metrics, indicating superior broad performance without excessive turnover.","kind":"performance","quote":"overall score (34), reflecting a combination of first‑ranked Sharpe performance and consistently strong risk metrics, particularly volatility and tail‑risk measures, without sacrificing turnover."},{"finding":"Statistical tests reject the null that several competitors (e.g., GS2, LS4) have Sharpe ratios equal to or higher than AIQ1, confirming AIQ1’s superiority at conventional significance levels.","kind":"statistical","quote":"the evidence supports rejection of H0 for the weaker methods, with GS2 and LS4 in particular exhibiting Sharpe ratios that are significantly lower than AIQ1 at conventional levels."}],"reported_performance":{"strategy":"Atomic‑IQ (AIQ1)","sharpe":0.56,"sharpe_benchmark":null,"annual_return_pct":5.91,"total_return_pct":null,"max_drawdown_pct":28.06},"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":39,"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":1,"extracted_at":"2026-11-14T17:17:45Z"},"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/"}