---
title: Covariance matrix filtering and portfolio optimisation: the average oracle vs non-linear shrinkage and all the variants of DCC-NLS
url: https://www.ml-quant.com/papers/arxiv/2309.17219/
site: ML-Quant (https://www.ml-quant.com)
updated: 2026-09-26
license: Summaries CC BY 4.0; links go to the original sources
index: https://www.ml-quant.com/llms.txt
identifier: arXiv:2309.17219
source_url: https://arxiv.org/abs/2309.17219
featured: 2023-10-04
citations: 6
topic: Portfolio & Allocation
---


# Covariance matrix filtering and portfolio optimisation: the average oracle vs non-linear shrinkage and all the variants of DCC-NLS

The Average Oracle, a fast covariance filtering method, outperforms complex methods, yielding superior Sharpe ratios in large-scale experiments.

- Source: https://arxiv.org/abs/2309.17219
- Identifier: arXiv:2309.17219
- Released: 2023-09-29
- First featured: Quant Letter No. 18 (2023-10-04): https://www.ml-quant.com/issues/2023-10-04/
- Citations (Semantic Scholar): 6
- Published in: Quantitative Finance
- Topic: Portfolio & Allocation

## Related

- [A Unified Framework for Fast Large-Scale Portfolio Optimization](https://www.ml-quant.com/papers/arxiv/2303.12751/): A new large-scale portfolio optimization framework, using shrinkage and regularization techniques, has been tested and proven effective using 50 years of US company return data.
- [Black-Litterman, Bayesian Shrinkage, and Factor Models in Portfolio Selection: You Can Have It All](https://www.ml-quant.com/papers/arxiv/2308.09264/): The paper introduces a Bayesian model that combines shrinkage estimation with view inclusion, applied to Fama-French approach factor models, outperforming simple and optimal portfolios based on sample estimators.
- [Shrinkage Estimator for Portfolios](https://www.ml-quant.com/papers/repec/gam-jjrfmx-v-15-y-2022-i-6-p-249-d-829749/): Study proposes shrinkage estimator for portfolio selection, outperforming traditional methods.
- [Consistent estimation of the high-dimensional efficient frontier](https://www.ml-quant.com/papers/arxiv/2409.15103/): The paper uses random matrix theory to study the asymptotic behavior of the mean-variance efficient frontier, finding that two out of three key quantities are biased and overestimated in high-dimensional asymptotic conditions.
- [The Critical Line Algorithm and the Constrained LASSO: One Curve, Two Literatures](https://www.ml-quant.com/papers/arxiv/2609.25704/): Shows that mean-variance portfolio selection and the constrained LASSO trace identical piecewise-linear solution paths, mapping their parametrizations exactly.
- [End-to-End, Decision-based, Cardinality-Constrained Portfolio Optimization](https://www.ml-quant.com/papers/ssrn/4598167/): The article introduces a comprehensive framework for portfolio optimization using a neural network, and compares three different problem-solving methods within this framework.