mean-variance-optimization

The paper proposes a single-level optimization formulation that incorporates KKT conditions into decision-focused learning for mean-variance portfolio optimization, improving performance in experiments on real-world ETF data.