Paper: arXiv 2611.00578

Authors: Yan Ding, Sotirios Sabanis, Ying Zhang

Abstract

Conditional-Value-at-Risk (CVaR) optimization is a fundamental tool in portfolio selection, yet its nonsmooth loss structure poses a challenge for the existing theory of stochastic gradient methods including associated Langevin algorithms. In particular, the positive-part loss and the softmax portfolio parametrization lead to stochastic gradients that need not be globally Lipschitz continuous, but rather discontinuous, where available non-convex stochastic gradient Langevin dynamics (SGLD) guarantees often exhibit an unfavorable exponential dependence on the problem dimension. In this article, we develop a non-asymptotic analysis of SGLD for a regularized CVaR portfolio optimization problem. We exploit the geometric structure of the CVaR objective to establish linear growth, convexity-at-infinity, and one-sided Lipschitz properties of the stochastic gradient without imposing density assumptions on the asset losses that are typical in previous works. Crucially, the constants appearing in the aforementioned structural conditions are shown to be independent of the number of assets. Moreover, we establish a convergence bound in Wasserstein-2 distance with rate $\mathcal{O}(λ^{1/4})$ and a finite-time guarantee for the expected excess risk. Our analysis, together with the dimension-independent constants in the structural conditions, leads to an excess risk bound that depends linearly on the number of assets. Numerical experiments with simulated Gaussian portfolios exhibit the predicted dimension dependence. Computational comparisons with alternative stochastic optimization methods further illustrate the practical efficiency of SGLD for the optimization problem under consideration. Finally, we extend the framework to a broader class of non-convex stochastic optimization problems with discontinuous stochastic gradients.

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 and theoretical analysis of SGLD for CVaR optimization, tackling a challenging problem with discontinuous gradients. While the empirical section is present, it is primarily illustrative with simulated data rather than extensive real-world backtesting. The novelty lies in the non-asymptotic analysis and dimension-independent constants.

Research Flowchart

  flowchart TD
    A[Research Goal: Non-asymptotic analysis of SGLD for CVaR minimization] --> B{Challenge: Discontinuous Stochastic Gradients in CVaR};
    B --> C[Methodology: SGLD with geometric analysis of CVaR objective];
    C --> D{Key Properties Identified: Linear growth, convexity-at-infinity, one-sided Lipschitz (dimension-independent)};
    D --> E[Computational Process: Numerical experiments with simulated Gaussian portfolios];
    E --> F[Key Outcome 1: Convergence bound in Wasserstein-2 distance O(λ^(1/4))];
    E --> G[Key Outcome 2: Excess risk bound linear in number of assets];
    F & G --> H[Broader Impact: Framework for non-convex optimization with discontinuous stochastic gradients];