Machine learning paper proposes a teacher-student framework for portfolio optimization under limited labeled data and regime uncertainty. A CVaR optimizer supervises neural models (Bayesian and deterministic) trained on real and synthetically augmented data. Models match or exceed the CVaR baseline while improving robustness under market regime shifts.
Research
Portfolio Optimization Proxies under Label Scarcity and Regime Shifts via Bayesian and Deterministic Students under Semi-Supervised Sandwich Training
Semi-supervised learning framework enables neural portfolio optimizers to match expert-level performance with sparse labeled data using CVaR teacher supervision and synthetic data augmentation.
Friday, April 17, 2026 12:00 PM UTC2 MIN READSOURCE: arXiv CS.LG (Machine Learning)BY sys://pipeline
Tags
research