← All quantitative finance open problems

Volatility models / Open question

Well-posedness of unregularized stochastic local volatility models

Provides a mathematical foundation for combining an observed vanilla-option surface with stochastic volatility dynamics used to price and hedge exotic derivatives.

OpenEditorially scopedFramework needs definitionCompiled 2026-11-09

Seed catalog · Showing the bundled record.

Problem statement

Establish sharp, verifiable conditions for existence and uniqueness of the unregularized calibrated dynamics dSt/St=σloc(t,St)Vt/E[Vt∣St] dWtdS_t/S_t=\sigma_{\mathrm{loc}}(t,S_t)\sqrt{V_t/\mathbb{E}[V_t\mid S_t]}\,dW_t, coupled to a continuous stochastic variance factor V. Determine when the law-dependent leverage coefficient is well defined and the stock-price marginals agree with those of the prescribed local-volatility model, including variance factors that can approach zero.

Assumptions and scope

  • Work under a risk-neutral measure with zero rates and dividends for the displayed equation; prescribe a positive regular local-volatility surface and initial joint law.
  • Specify the continuous variance-factor SDE, correlation, coefficient growth and boundary conditions; distinguish weak existence, uniqueness in law and pathwise uniqueness.
  • Interpret conditional expectation through the joint law and require positivity wherever the denominator is used. A fixed positive denominator floor or kernel smoothing changes the problem.
  • Impose the integrability and true-martingale conditions needed to identify European option prices from the marginals.

What would settle it?

A well-posedness and marginal-calibration theorem for a clearly specified class with continuous, possibly degenerate variance factors, or counterexamples locating the necessary restrictions; if regularization is used, justify its removal.

Selected for a theoretical or computational target. A successful calibration or numerical benchmark alone does not establish the stated mathematical guarantee.

Status and evidence

Regularized equations have well-posedness and particle-convergence results. Mustapha's 2024 strong-existence/uniqueness result assumes a finitely valued volatility factor; other existence and stationary special-case results also exist. The remaining target is a sharp theory for unregularized continuous-factor models, not a claim that all LSV models lack solutions. Source abstracts were checked; no exhaustive 2026 status certification is asserted.

Preliminary source check; not an expert-verified open status or an independent audit of proposed proofs. The compilation date is not a literature-review cutoff.

Sources and further reading

  1. Bayer, Belomestny, Butkovsky & Schoenmakers: A reproducing kernel Hilbert space approach to singular local stochastic volatility McKean–Vlasov models ↗partial progress · source checked 2026-11-09

    Scholarly abstract: regularization, well-posedness and propagation of chaos.

    Explicitly distinguishes the unresolved original equation from the regularized model.

  2. Reisinger & Tsianni: Convergence of the Euler–Maruyama particle scheme for a regularised McKean–Vlasov equation arising from the calibration of local-stochastic volatility models ↗partial progress · source checked 2026-11-09

    Scholarly abstract: strong convergence with fixed regularization parameters.

    Convergence for a regularized scheme does not establish the zero-regularization limit.

  3. Mustapha: Strong existence and uniqueness of a calibrated local stochastic volatility model ↗partial progress · source checked 2026-11-09

    arXiv abstract: finitely valued factor assumption.

    A solved restricted model; the title should not be read as a general continuous-factor theorem.