Technology Category
Machine Learning: OptimizationConstraint Satisfaction and Optimization: Constraint OptimizationSearch and Optimization: Non-convex Optimization
Multi-objective Portfolio Optimization Via Gradient Descent
🤖 AI Summary
Traditional portfolio optimization methods face scalability limitations and insufficient modeling flexibility when handling multi-objective formulations, complex regulatory constraints (e.g., UCITS), tracking-error bounds, and large-scale datasets. To address these challenges, we propose an end-to-end multi-objective optimization framework grounded in gradient descent and automatic differentiation. The framework supports arbitrary risk–return objectives—such as CVaR and Sharpe ratio—in flexible combinations, and unifies hard and soft constraints within a single differentiable computational graph. It enables seamless transitions between single- and multi-objective settings without architectural modification. Empirically evaluated across six realistic scenarios, our approach matches or exceeds the performance of state-of-the-art solvers—including CVXPY and SKFOLIO—while substantially improving modeling expressiveness, computational scalability, and practical deployability. This work establishes a new paradigm for modern asset allocation that bridges theoretical rigor with engineering feasibility.
Technology Category
Application Category
📝 Abstract
Traditional approaches to portfolio optimization, often rooted in Modern Portfolio Theory and solved via quadratic programming or evolutionary algorithms, struggle with scalability or flexibility, especially in scenarios involving complex constraints, large datasets and/or multiple conflicting objectives. To address these challenges, we introduce a benchmark framework for multi-objective portfolio optimization (MPO) using gradient descent with automatic differentiation. Our method supports any optimization objective, such as minimizing risk measures (e.g., CVaR) or maximizing Sharpe ratio, along with realistic constraints, such as tracking error limits, UCITS regulations, or asset group restrictions. We have evaluated our framework across six experimental scenarios, from single-objective setups to complex multi-objective cases, and have compared its performance against standard solvers like CVXPY and SKFOLIO. Our results show that our method achieves competitive performance while offering enhanced flexibility for modeling multiple objectives and constraints. We aim to provide a practical and extensible tool for researchers and practitioners exploring advanced portfolio optimization problems in real-world conditions.
Problem
Research questions and friction points this paper is trying to address.
Address scalability and flexibility in portfolio optimization
Support diverse objectives and realistic constraints
Compare performance with standard solvers in complex scenarios
Innovation
Methods, ideas, or system contributions that make the work stand out.
Gradient descent with automatic differentiation
Supports any optimization objective and constraints
Competitive performance in multi-objective scenarios
🔎 Similar Papers
No similar papers found.
