Autonomous Sparse Mean-CVaR Portfolio Optimization

Paper: arXiv 2405.08047

Abstract

The $\ell_0$-constrained mean-CVaR model poses a significant challenge due to its NP-hard nature, typically tackled through combinatorial methods characterized by high computational demands. From a markedly different perspective, we propose an innovative autonomous sparse mean-CVaR portfolio model, capable of approximating the original $\ell_0$-constrained mean-CVaR model with arbitrary accuracy. The core idea is to convert the $\ell_0$ constraint into an indicator function and subsequently handle it through a tailed approximation. We then propose a proximal alternating linearized minimization algorithm, coupled with a nested fixed-point proximity algorithm (both convergent), to iteratively solve the model. Autonomy in sparsity refers to retaining a significant portion of assets within the selected asset pool during adjustments in pool size. Consequently, our framework offers a theoretically guaranteed approximation of the $\ell_0$-constrained mean-CVaR model, improving computational efficiency while providing a robust asset selection scheme.

Complexity vs Empirical Score

• Math Complexity: 8.5/10
• Empirical Rigor: 3.0/10
• Quadrant: Lab Rats — theoretically deep, empirically untested

Why this score: The paper introduces advanced mathematical concepts such as an NP-hard ℓ0-constrained mean-CVaR model, a novel tailed approximation for the indicator function, and a convergent PALM algorithm with nested FPPA, indicating high mathematical density. However, the summary and excerpt focus on theoretical contributions (approximation guarantees, convergence proofs) with no mention of backtesting, datasets, or implementation details, suggesting low empirical rigor.

Research Flowchart

  flowchart TD
  A["Research Goal"] --> B["Core Idea"]
  B --> C["Algorithm Development"]
  C --> D["Experimental Validation"]
  D --> E["Key Outcomes"]

  subgraph A ["Research Goal"]
      A1["Approximate l₀-constrained<br>Mean-CVaR Model"]
  end

  subgraph B ["Core Methodology"]
      B1["Convert l₀ to Indicator Function"]
      B2["Tailed Approximation"]
      B3["Autonomous Sparsity Control"]
  end

  subgraph C ["Computational Process"]
      C1["Proximal Alternating<br>Linearized Minimization"]
      C2["Nested Fixed-Point<br>Proximity Algorithm"]
  end

  subgraph D ["Data & Inputs"]
      D1["Historical Asset Returns"]
      D2["Portfolio Constraints"]
  end

  subgraph E ["Outcomes"]
      E1["Theoretical Guarantee<br>of Approximation"]
      E2["Improved Computational<br>Efficiency"]
      E3["Robust Asset Selection<br>Scheme"]
  end

Cited References (126)

1. Multitrend Conditional Value at Risk for Portfolio Optimization (2022) — link
2. Perspective reformulations of mixed integer nonlinear programs with indicator variables (2010)
3. On cardinality constrained mean-CVaR portfolio optimization (2015)
4. Bilevel cutting-plane algorithm for cardinality-constrained mean-CVaR portfolio optimization (2021)
5. A Scalable Algorithm for Sparse Portfolio Selection (2022)
6. Coherent Measures of Risk (1999)
7. Approaching Mean-Variance Efficiency for Large Portfolios (2018)
8. Improving Mean Variance Optimization through Sparse Hedging Restrictions (2015)

0 of 126 cited works are in the corpus; resolved via arXiv source bundles, Semantic Scholar and OpenAlex. Corpus-wide: 2518 linked papers.