{"paper_id":"2506.10817","title":"On the Weak Error for Local Stochastic Volatility Models","authors":["Peter K. Friz","Benjamin Jourdain","Thomas Wagenhofer","Alexandre Zhou"],"paper_date":"2025-06-12","paper_type":"theoretical","paper_url":"https://arxiv.org/pdf/2506.10817v1","page_url":"https://thequant.space/flowcharts/on-the-weak-error-for-local-stochastic-volatility-models/","scores":{"empirical_rigor":4,"math_complexity":9,"quadrant":"Lab Rats"},"proposes":{"status":"available","statement":"The paper proposes a well‑defined Euler approximation to the formal McKean‑Vlasov equation together with a newly established half‑step scheme for approximating conditional expectations, and a particle system to implement the approximation.","quote":"Our take is to start with a well-defined Euler approximation to the formal McKean-Vlasov equation, followed by a newly established half-step-scheme, allowing for good approximations of conditional expectations. We furthermore formally introduce the half-step scheme.","summary":"Local stochastic volatility refers to a popular model class in applied mathematical finance that allows for “calibration-on-the-fly”, typically via a particle method, derived from a formal McKean-Vlasov equation. Well-posedness of this limit is a well-known problem in the field; the general case is…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The stochastic volatility variable ξ is bounded from above and below.","type":"model","stated":"explicit","quote":"he boundedness of ξ and the assumptions 10 on bE[ξ|·] it follows that with hidden constants depending on p > 1 1 2 E h\u0010 ξ2 jh bE \u0002 ξ2 jh b Xh jh \u0003 − ξ2 jh E \u0002 ξ2 jh b Xh jh \u0003 \u0011 σ2\u0000jh, b Xh jh \u0001 ∂2 xu \u0000jh, b Xh jh \u0001i ≲E h\u0010 1 bE \u0002 ξ2 jh b Xh jh \u0003 − 1 E \u0002 ξ2 jh b Xh jh \u0003 \u0011pi1/p ≲E h\u0010 bE \u0002 ξ2 jh b Xh jh \u0003 −E \u0002 ξ2 jh b Xh jh \u0003\u0011pi1/p .","support":"full"},{"assumption":"The drift and diffusion coefficients satisfy polynomial growth conditions and have bounded moments.","type":"model","stated":"explicit","quote":"By polynomial growth condition for f ′ and existence of moments for Y , see Lemma 9.3 with ¯β ≡0, there is some N ∈N, C > 0 such that u(T, x) −u(T −h, x) = E h f(x) −f \u0000Y T −h,x T ) i ≤CE h |x −Y T −h,x T | \u00001 + |x| + |Y T −h,x T | \u0001Ni ≲ √ h(1 + |x|)N. 6 Here we used H¨older inequality and the fact that ∥x −Y T −h,x T ∥Lp ≲ √ h for all p ≥1.","support":"full"},{"assumption":"Test functions f are three times weakly differentiable with derivatives of polynomial growth.","type":"statistical","stated":"explicit","quote":"Let h, T be such that T h ∈N and f : R →R be three times weakly differentiable with derivatives of polynomial growth.","support":"full"},{"assumption":"The regularisation parameter δ lies in (0, 1/2) and the kernel estimator satisfies Lipschitz properties.","type":"model","stated":"explicit","quote":"In this section we show that the contribution to the weak error from this additional δ-parameter can be well-quantified. 13 Theorem 4.6. Under Assumptions 1 and 2, for e Xh,δ and b Xh,δ respectively defined in (14) and (17) it holds for any p > 1 and δ ∈(0, 1/2) that E \u0002 f(YT ) \u0003 −E \u0002 f \u0000 e Xh,δ T \u0001\u0003 = E \u0002 f(YT ) \u0003 −E \u0002 f \u0000 b Xh,δ T \u0001\u0003 ≲h + δ 1 p Here the hidden constant also depends on p. The proof of this th","support":"full"},{"assumption":"The particle system satisfies a propagation of chaos result under Assumptions 1 and 2.","type":"model","stated":"explicit","quote":"For this we introduce a particle system and show a propagation of chaos result, following a well-established method, see e.g. [20, Proposition 4.1]. Let λ, δ ∈(0, 1) and µ ∈P2(R × R) induced by some random variables (ξ, X) such that ξ is bounded from above and below as in Assumption 2.","support":"full"},{"assumption":"The half‑step scheme uses parameters ξ, σ, δ, h, ρ, c_min and N as defined in the paper.","type":"other","stated":"explicit","quote":"The parameters used are • ξ for the stochastic volatility, • σ for the local volatility, • δ for regularisation, • h for the discretisation length, • ρ for the correlation between B and W • cmin for the lower bound in Assumption 2 • N for the number of particles, which is equal to the number of Monte Carlo steps.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Rough‑Bergomi‑type stochastic volatility model","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"For the stochastic volatility we chose a rough-Bergomi-type model, i.e. ξt = 0.01 + 0.5 exp \u0000W H t −t2H 2 \u0001 , with cmin = 0.051 and where W H t denotes a Riemann-Liouville fractional Brownian motion with Hurst parameter H = 0.1, fully correlated to W. Recall that W H t = CH R t 0(t−s)H−1/2 dWs, where CH is chosen such that E[(W H t )2] = t2h for all t ∈[","support":"full"},{"name":"Tanh volatility model σ(x)=3/4+ tanh(x)/4","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"6.2 Tanh model In this section we use a non-trivial volatility, namely we choose σ(x) = 3 4 + tanh(x) 4 .","support":"full"}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"synthetic"},"transaction_costs":{"status":"not_applicable","modeled":"n/a","headline_net_of_costs":null,"detail":null,"quote":null,"source":"front_matter","rule":"R3","candidates":{"front_matter":"n/a","evidence_card":"n/a","evidence_card_quoted":false},"meaning":"the paper reports no trading or execution result"},"results":{"status":"available","evaluation_type":"unclear","findings":[{"finding":"The Euler discretization achieves weak order one, with error terms accounting for the particle approximation.","kind":"theoretical","quote":"We show weak order one for the Euler discretization, plus error terms that account for the said approximation.","support":"full"},{"finding":"The weak error of the particle scheme is quantified in dependence on all parameters (h, δ, N, ε, etc.).","kind":"theoretical","quote":"The case of particle approximation is discussed in detail and the error rate is given in dependence of all parameters used. Similar to Theorem 1, Theorem 7.2 states the weak error depending on all the parameters used.","support":"full"},{"finding":"Numerical experiments show that the error does not strongly depend on the step size.","kind":"performance","quote":"It is also very remarkable that the error does not really depend on the step size.","support":"full"},{"finding":"Stability of the scheme is achieved already at a step size of approximately h≈0.1.","kind":"performance","quote":"Stability was already achieved at a step size of h ≈0.1.","support":"full"},{"finding":"For the smooth test function the absolute error is very small due to the large number of particles.","kind":"performance","quote":"Note that due to the relatively big number of particles and the smooth test function, the absolute value of the error is quite small.","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":16,"citations":{"cited_by_in_corpus":0,"by_independent_authors":0,"how_cited":{"used_as_baseline":0,"uses_method_or_data":0,"extends":0,"discusses_findings":0,"mentions":0,"unclassified":0},"independent_papers":[],"scope":"Citations between papers in this corpus only; citations from outside it are not counted."}},"implementation":{"status":"available","public":false,"code_url":null,"note":"No repository presented by the authors as their own was found in the paper. Absence here is not proof that no code exists."},"replication":{"status":"none_found","card":null,"note":"No replication of this paper has been run by thequant.space yet."},"extraction":{"text_source":"fulltext","model":"openai/gpt-oss-120b","version":2,"extracted_at":"2026-11-14T19:49:50Z"},"provenance":{"claims":"Everything here is what the authors state in the paper. Nothing was re-run or reproduced by the platform.","quotes":"Assumptions, datasets, cost detail and findings were extracted by a language model. Every quote is verbatim paper text. Where an item has `support`, \"full\" means every figure in it appears in the quoted sentences and \"partial\" means a figure is stated elsewhere in the paper; items with a figure found nowhere in the paper were discarded.","datasets_known":"datasets.known lists well-known vendors and public datasets by a stable id, found by a dictionary over the full text where the sentence says the data is used; it is independent of the model and covers only names in the dictionary.","flags":"transaction_costs.modeled, results.evaluation_type and results.out_of_sample.method are platform classifications (structured-decision models); see https://thequant.space/score-guide/#penalty-flags","comparisons":"supports / contradicts / qualifies are a language model's reading of two papers' quoted findings, restricted to closely related papers that share no author. They are pointers to read, not verified replications, and coverage is partial.","implementation":"code_url is a repository the paper itself presents as its own (rule-scored, ambiguous cases judged by a model), checked for liveness when it was recorded."},"docs":"https://thequant.space/agents/"}