Pith. sign in

REVIEW 2 cited by

A fully data-driven approach to minimizing CVaR for portfolio of assets via SGLD with discontinuous updating

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2007.01672 v1 pith:3SYZKWIA submitted 2020-07-02 q-fin.PM math.OCmath.PRmath.STq-fin.MFstat.TH

A fully data-driven approach to minimizing CVaR for portfolio of assets via SGLD with discontinuous updating

classification q-fin.PM math.OCmath.PRmath.STq-fin.MFstat.TH
keywords stochasticapproachsgldassetscvardata-drivendiscontinuousfindings
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

A new approach in stochastic optimization via the use of stochastic gradient Langevin dynamics (SGLD) algorithms, which is a variant of stochastic gradient decent (SGD) methods, allows us to efficiently approximate global minimizers of possibly complicated, high-dimensional landscapes. With this in mind, we extend here the non-asymptotic analysis of SGLD to the case of discontinuous stochastic gradients. We are thus able to provide theoretical guarantees for the algorithm's convergence in (standard) Wasserstein distances for both convex and non-convex objective functions. We also provide explicit upper estimates of the expected excess risk associated with the approximation of global minimizers of these objective functions. All these findings allow us to devise and present a fully data-driven approach for the optimal allocation of weights for the minimization of CVaR of portfolio of assets with complete theoretical guarantees for its performance. Numerical results illustrate our main findings.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 2 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Tamed Stochastic Gradient Hamiltonian Monte Carlo

    math.OC 2026-07 conditional novelty 5.0

    tSGHMC provably samples from strongly convex targets with superlinear, discontinuous stochastic gradients at a λ^{1/4} Wasserstein-2 rate.

  2. Robust SGLD algorithm for solving non-convex distributionally robust optimisation problems

    math.OC 2024-03 unverdicted novelty 5.0

    Develops robust SGLD with non-asymptotic convergence bounds for non-convex DRO and applies it to neural network regression under adversarial corruption.