Decision-Induced Ranking Explains Prediction Inflation and Excessive Turnover in SPO-Based Portfolio Optimization

2026/05/02 by Yi Wang, Takashi Hasuike · 1 voice
Economics, Econometrics and Finance · #q-fin.PM #q-fin.CP

paper · pdf

arxiv published 2026/05/02 · arxiv updated 2026/07/13

Abstract

Decision-focused learning (DFL) is attractive for portfolio optimization because it trains predictors according to downstream decision quality rather than prediction accuracy alone. However, SPO(Smart, Predict then Optimize surrogate)-based DFL may produce inflated return signals and unstable portfolio reallocations. This study provides a KKT-based interpretation showing that portfolio decisions can be viewed as ranking over risk- and transaction-cost-adjusted marginal scores. Empirically, we examine prediction inflation and excessive turnover in SPO-trained portfolios, and evaluate clipping, min-max rescaling, and partial portfolio adjustment as practical stabilization mechanisms. The results suggest that realistic output constraints and portfolio-level turnover control improve the implementability of SPO-based portfolio strategies.

Citations

• Estimating Covariance for Global Minimum Variance Portfolio: A Decision-Focused Learning Approach
• Return Prediction for Mean-Variance Portfolio Selection: How Decision-Focused Learning Shapes Forecasting Models
• Distributionally Robust End-to-End Portfolio Construction
• Decision-Focused Learning: Through the Lens of Learning to Rank
• Integrating prediction in mean-variance portfolio optimization
• PORTFOLIO SELECTION*

Discussions

• Yi Wang, Takashi Hasuike: Decision-Induced Ranking Explains Prediction Inflation and Excessive Turnover in SPO-Based Portfolio Optimization https://arxiv.org/abs/2605.01176 https://arxiv.org/pdf/2605. [bsky, 0 points, 0 comments]

Related

• Smart Predict--then--Optimize Paradigm for Portfolio Optimization in Real Markets
• Return Prediction for Mean-Variance Portfolio Selection: How Decision-Focused Learning Shapes Forecasting Models
• Integrated Prediction and Multi-period Portfolio Optimization
• DFF: Decision-Focused Fine-tuning for Smarter Predict-then-Optimize with Limited Data
• Decision-informed Neural Networks with Large Language Model Integration for Portfolio Optimization