End-to-End Neural Shrinkage of Indefinite Pairwise Correlation Matrices for Small-Cap-Inclusive Portfolios

arXiv:2608.30446 · q-fin.PM, cs.LG, q-fin.ST · Submitted 2026-08-31 · Read on arXiv

q-fin.PM, cs.LG, q-fin.ST

Submitted: 2026-08-31

Updated: 2026-08-31

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Benchmark — 5× in this paper · explained in 2 episodes
• Model Confidence Set — 4× in this paper
• Regularization — 2× in this paper · explained in 3 episodes
• Adam — 1× in this paper · explained in 2 episodes
• Existing Random Matrix Theory — 1× in this paper
• Global Minimum Variance — 1× in this paper
• MLE and Quadratic Inverse Shrinkage — 1× in this paper
• Net Liquidation Value — 1× in this paper
• Provenance — 1× in this paper · explained in 3 episodes
• Random Matrix Theory (RMT) — 1× in this paper · explained in 3 episodes
• alpha — 1× in this paper · explained in 1 episode

Sources

• On estimating covariances between many assets with histories of highly variable length

Related papers

• Large Language Model-Driven Small-Capitalization Trading: Integrating Financial News Sentiment, Macroeconomic Indicators, and Technical Signals
• FactorBench: A Portfolio-Aware Benchmark for Automated Factor Mining
• Taming the Greeks: Option Portfolios with Inductive Biases
• Financially Guided Deep Portfolio Optimization
• Eliciting ESG Preferences for Reinforcement Learning-Based Portfolio Optimization
• Data-Driven Duration Management -- Term Structure Forecasting Using Machine Learning