{"paper_id":"2611.25344","title":"Quantum Calibration of Local-Volatility Surfaces from Sparse Option Quotes","authors":["Kentaro Hoshisashi"],"paper_date":"2026-11-19","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2611.25344","page_url":"https://thequant.space/flowcharts/quantum-calibration-of-local-volatility-surfaces-from-sparse-option-quotes/","scores":{"empirical_rigor":7,"math_complexity":8.5,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes a hybrid quantum‑classical calibration method for the Dupire local‑volatility model, using a positive coefficient representation of local variance and linear combinations of Hamiltonian simulations to approximate pricing evolution, with measurement counts feeding a classical optim","quote":"We implement a hybrid quantum and classical method for this inverse problem using the Dupire local-volatility model. A positive coefficient representation specifies the local variance, and linear combinations of Hamiltonian simulations approximate the associated pricing evolution.","summary":"Calibrating a volatility surface from sparse option prices requires repeated solutions of a pricing equation as the volatility parameters change. We implement a hybrid quantum and classical method for this inverse problem using the Dupire local-volatility model. A positive coefficient…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"Local variance is non‑negative, enforced by the positive coefficient representation.","type":"model","stated":"explicit","quote":"On interval s, we use as(k; ξs) = amin + M X j=1 ξsjϕj(log k), 0 ≤ξsj ≤1, (3) where ϕj ≥0 and amin = 0.01. Positivity is imposed by the representation.","support":"full"},{"assumption":"Basis functions ϕj are non‑negative and control different parts of the local‑variance curve.","type":"model","stated":"explicit","quote":"On interval s, we use as(k; ξs) = amin + M X j=1 ξsjϕj(log k), 0 ≤ξsj ≤1, (3) where ϕj ≥0 and amin = 0.01. Positivity is imposed by the representation.","support":"full"},{"assumption":"Linear interpolation in strike preserves strike‑convexity conditions between grid nodes.","type":"model","stated":"explicit","quote":"Linear strike interpolation preserves the strike conditions between nodes.","support":"full"},{"assumption":"Exact propagation and backward‑Euler discretisation preserve the discrete inequalities for positive local variance.","type":"model","stated":"explicit","quote":"For positive local variance, the fitted spatial operator pre-serves the corresponding discrete inequalities under exact propagation and backward Euler.","support":"full"},{"assumption":"Observed strikes may not coincide with grid nodes, and linear interpolation is used to map observations.","type":"data","stated":"explicit","quote":"Observed strikes need not coincide with grid nodes. We use linear interpolation in strike, including the boundary values, to define the full observation map W full s .","support":"full"},{"assumption":"Synthetic examples use a positive affine representation with three basis functions and specified coefficient values.","type":"other","stated":"explicit","quote":"Both use the positive affine representation in (3), with three basis functions of x = log k ϕ1(x) = 0.08, ϕ2(x) = 0.1(x + 0.45)2, ϕ3(x) = 0.1(x −0.45)2. (13) The first coefficient controls the variance level.","support":"full"}]},"datasets":{"status":"available","items":[{"name":"Synthetic smile and skew examples","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"EXPERIMENTAL DESIGN A. Surface Reconstruction We introduce two synthetic examples to show the financial objects produced by the framework: a smile and a negatively sloped skew. Both use the positive affine representation in (3), with three basis functions of x = log k ϕ1(x) = 0.08, ϕ2(x) = 0.1(x + 0.45)2, ϕ3(x) = 0.1(x −0.45)2.","support":"full"},{"name":"Benchmark calibration dataset (clean and noisy quotes)","provider":null,"kind":"options","access":"unknown","period":null,"universe":null,"quote":"B. Classical Comparators The larger benchmark contains independent calibrations using matrix exponentiation, sparse backward Euler and two LCHS kernels within the same four-basis model. Clean data and uniform price perturbations bounded by 5 basis points (bp) are considered, with three noise datasets for the latter.","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":"pricing-calibration","findings":[{"finding":"At the common fitted coefficients, exact and LCHS held‑out pricing errors are 0.27 bp for the smile and 0.26 bp for the skew.","kind":"pricing-calibration","quote":"At the common fitted coefficients, exact and LCHS held-out errors are 0.27 and 0.27 bp for the smile, and 0.26 and 0.26 bp for the skew.","support":"full"},{"finding":"Beta LCHS stays within 0.03 bp of exact propagation, while finite‑difference stays within the same model error range.","kind":"pricing-calibration","quote":"Beta LCHS stays within 0.03 bp of exact propagation and finite differences (FD) within the same model.","support":"full"},{"finding":"Doubling the reference grid from 1,024 to 2,048 nodes changes quote prices by less than 0.01 bp RMSE in each synthetic example.","kind":"performance","quote":"Doubling the reference grid from 1,024 to 2,048 nodes changes quote prices by less than 0.01 bp in RMSE in each example.","support":"full"},{"finding":"Increasing repetition budgets lowers the mean price‑error of the final residual measurement, but three seeds show no clear trend.","kind":"statistical","quote":"More repetitions lower the mean here, though three seeds establish no trend and finite sampling also changes the sequence of selected candidates. They are appreciably wider at the lower repetition budget.","support":"full"},{"finding":"Increasing the LCHS term count reduces price error and eliminates negative in‑band butterfly prices in a stress case, though violations remain outside the band.","kind":"performance","quote":"Fixed-coefficient stress case. More LCHS terms reduce price error (top) and remove negative in-band butterflies (bottom).","support":"full"}],"reported_performance":null,"out_of_sample":{"method":"unclear","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":22,"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":true,"code_url":"https://github.com/khoshisashi/quantum-local-vol","note":null},"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":3,"extracted_at":"2026-11-20T11:38:35Z"},"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/"}