Paper: arXiv 2610.13162

Authors: Xinagyu Cui, Xun Li, Ziyan Li, Yun Shi

Abstract

This paper studies entropy regularization for the discrete-time dynamic mean–variance portfolio problem with no-shorting constraints. In the unconstrained setting, entropy regularization can be introduced without changing the basic quadratic structure of the dynamic program. Under no-shorting, this property is lost: a standard fixed-temperature formulation generally preserves convexity but no longer preserves the homogeneous Bellman structure and the associated piecewise-linear feedback form. We introduce a state-dependent temperature that is matched to the homogeneity of the original mean–variance problem. This restores the piecewise-quadratic value function and the finite-dimensional Bellman recursion. The resulting formulation, however, makes convexity depend on the continuation coefficients generated by the soft recursion. We show that negative continuation coefficients lead to loss of convexity and prove that a common sufficiently small temperature preserves the Bellman structure and convexity over the entire finite horizon. The original no-shorting solution is recovered as the temperature vanishes. We also develop a sample-based implementation for unknown return distributions. Numerical and empirical results illustrate both the structural advantage of the state-dependent temperature and the practical relevance of the convexity-preserving temperature restriction.

Complexity vs Empirical Score

  • Math Complexity: 9.0/10
  • Empirical Rigor: 6.0/10
  • Quadrant: Holy Grail — high math complexity, high empirical rigor

Why this score: The paper presents a highly mathematical treatment of entropy regularization in dynamic mean-variance optimization, introducing a novel state-dependent temperature to preserve structure. While primarily theoretical, it includes numerical and empirical results, demonstrating a solid level of empirical rigor in validating the proposed methodology. The combination of advanced mathematical derivations and practical application places it in the ‘Holy Grail’ quadrant.

Research Flowchart

  flowchart TD
    A[Research Goal: Address entropy regularization for dynamic mean-variance portfolio with no-shorting] --> B{Methodology: Entropy Regularization & State-Dependent Temperature};
    B --> C[Problem Formulation: Standard vs. State-Dependent Temperature];
    C --> D{Analysis & Derivations: Bellman Equations, Convexity, Structure Preservation};
    D --> E[Computational Processes: Numerical Simulations & Empirical Studies];
    E --> F{Key Outcomes: State-Dependent Temperature Restores Structure & Convexity under Sufficiently Small Temperature};
    F --> G[Further Outcomes: Original No-Shorting Solution Recovered as Temperature Vanishes; Sample-based Implementation Developed];

Cited References (65)

  1. Portfolio Selection (1952)
  2. Optimal Dynamic Portfolio Selection: Multiperiod Mean-Variance Formulation (2000)
  3. Dynamic Mean-Variance Asset Allocation (2010)
  4. Survey on Multi-Period Mean–Variance Portfolio Selection Model (2022)
  5. Dynamic Mean-Variance Portfolio Selection with No-Shorting Constraints (2002)
  6. Optimal Multi-Period Mean–Variance Policy under No-Shorting Constraint (2014)
  7. Mean-Variance Policy for Discrete-Time Cone-Constrained Markets: Time Consistency in Efficiency and the Minimum-Variance Signed Supermartingale Measure (2017)
  8. Dynamic Factor Model-Based Multiperiod Mean-Variance Portfolio Selection with Portfolio Constraints — → 2502.17915
and 57 more
  1. Time-Consistent Portfolio Selection under Short-Selling Prohibition: From Discrete to Continuous Setting (2014)
  2. The Markowitz Optimization Enigma: Is ``Optimized’’ Optimal? (1989)
  3. Estimation for Markowitz Efficient Portfolios (1980)
  4. Mean–Variance Portfolio Optimization When Means and Covariances Are Unknown (2011)
  5. Optimal Portfolio Choice with Parameter Uncertainty (2007)
  6. Integrated Conditional Estimation-Optimization (2026)
  7. Estimate-then-Optimize versus Integrated-Estimation-Optimization versus Sample Average Approximation: A Stochastic Dominance Perspective (2023) — link
  8. Dissecting the Impact of Model Misspecification in Data-Driven Optimization (2025) — link
  9. Toward Decision-Oriented Prognostics: An Integrated Estimate-Optimize Framework for Predictive Maintenance (2025) — link
  10. Task-Based End-to-End Model Learning in Stochastic Optimization (2017)
  11. Smart ``Predict, then Optimize’’ (2022)
  12. Melding the Data-Decisions Pipeline: Decision-Focused Learning for Combinatorial Optimization (2019)
  13. From Predictive to Prescriptive Analytics (2020)
  14. Integrated Prediction and Multi-Period Portfolio Optimization (2025) — link
  15. Reinforcement Learning in Continuous Time and Space: A Stochastic Control Approach (2020)
  16. Continuous-Time Mean–Variance Portfolio Selection: A Reinforcement Learning Framework (2020)
  17. Discrete-Time Mean-Variance Strategy Based on Reinforcement Learning — → 2312.15385
  18. Soft Actor-Critic: Off-Policy Maximum Entropy Deep Reinforcement Learning with a Stochastic Actor (2018)
  19. Soft Actor-Critic Algorithms and Applications (2018) — link
  20. Asymptotic Evaluation of Certain Markov Process Expectations for Large Time. IV (1983)
  21. A Weak Convergence Approach to the Theory of Large Deviations (2011)
  22. Elements of Information Theory (1999)
  23. Reinforcement Learning and Control as Probabilistic Inference: Tutorial and Review (2018) — link
  24. Variational Characterization of Free Energy: Theory and Algorithms (2017)
  25. Large Deviations Techniques and Applications (2009)
  26. Bayesian Inference in Econometric Models Using Monte Carlo Integration (1989)
  27. Distribution of Points in a Cube and Approximate Evaluation of Integrals (1967)
  28. Monte Carlo Variance of Scrambled Net Quadrature (1997)
  29. Randomized Quasi-Monte Carlo: An Introduction for Practitioners (2016)
  30. Numerical Computation of Multivariate Normal Probabilities Using Bivariate Conditioning (2016)
  31. The Normal Law under Linear Restrictions: Simulation and Estimation via Minimax Tilting (2017)
  32. Policy Learning and Evaluation with Randomized Quasi-Monte Carlo (2022) — link
  33. Finite Mixture Models (2000)
  34. Estimating the Dimension of a Model (1978)
  35. Time Series Analysis (2020)
  36. Common Risk Factors in the Returns on Stocks and Bonds (1993)
  37. Policy Optimization Reinforcement Learning with Entropy Regularization (2019) — link
  38. An Adaptive Entropy-Regularization Framework for Multi-Agent Reinforcement Learning (2023)
  39. Soft Actor-Critic with Automatically Adjusted Entropy for Autonomous Exploring in Unknown Environments (2025)
  40. Modeling Purposeful Adaptive Behavior with the Principle of Maximum Causal Entropy (2010)
  41. Maximum Entropy Reinforcement Learning with Diffusion Policy (2025) — link
  42. Sparse Markov Decision Processes with Causal Sparse Tsallis Entropy Regularization for Reinforcement Learning (2018)
  43. Machine Learning and Portfolio Optimization (2018) — link
  44. Training with Noise is Equivalent to Tikhonov Regularization (1995)
  45. Robustness and Regularization of Support Vector Machines (2009)
  46. Dynamic Mean–Variance Portfolio Selection with No-Shorting Constraints and Unknown Investment Opportunity Sets (2026) — link
  47. Portfolio Constraints: An Empirical Analysis (2022) — link
  48. Continuous-Time Mean–Variance Portfolio Selection: A Stochastic LQ Framework (2000) — link
  49. Cone-Constrained Continuous-Time Markowitz Problems (2013) — link
  50. Reinforcement Learning in Continuous Time and Space: A Stochastic Control Approach (2020)
  51. Exploratory HJB Equations and Their Convergence (2022) — link
  52. Learning Equilibrium Mean–Variance Strategy (2023) — link
  53. Optimal Scheduling of Entropy Regularizer for Continuous-Time Linear-Quadratic Reinforcement Learning (2024) — link
  54. Reinforcement Learning for Continuous-Time Mean–Variance Portfolio Selection in a Regime-Switching Market (2024) — link
  55. Continuous-Time Mean–Variance Portfolio Selection: A Reinforcement Learning Framework (2020) — link
  56. Discrete-Time Mean-Variance Strategy Based on Reinforcement Learning — → 2312.15385
  57. Entropy-Regularized Mean-Variance Portfolio Optimization with Jumps (2023)

3 of 65 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.