A decision-focused learning framework for goal-based investing

Hyunglip Bae, Minsu Park, Haeun Jeon, Woo Chang Kim

Abstract: This paper explores the application of Decision-focused learning (DFL) within the framework of Goal-Based Investing (GBI), highlighting its advantages over traditional Prediction-focused Learning (PFL) methods. A key contribution of this research is the introduction of a novel decision quality metric tailored for goal programming, prioritizing investor-specific goals in multi-objective optimization. Our findings demonstrate that DFL enhances decision quality and portfolio feasibility, particularly under uncertain market conditions, outperforming PFL in aligning investment strategies with prioritized financial goals. By integrating decision-making directly into the learning process, DFL enables more accurate parameter predictions, resulting in improved portfolio performance. This study underscores DFL's practical effectiveness in navigating market complexities and its potential as a powerful tool for optimizing portfolios in alignment with long-term investment objectives.