{"paper_id":"2507.13186","title":"NUFFT for the Fast COS Method","authors":["Fabien LeFloc’h"],"paper_date":"2025-07-17","paper_type":"empirical","paper_url":"https://arxiv.org/pdf/2507.13186v2","page_url":"https://thequant.space/flowcharts/nufft-for-the-fast-cos-method/","scores":{"empirical_rigor":7.8,"math_complexity":9.2,"quadrant":"Holy Grail"},"proposes":{"status":"available","statement":"The paper proposes adapting the COS option pricing method to use a type‑2 non‑uniform fast Fourier transform, achieving unprecedented speed for pricing many options of the same maturity but different strikes.","quote":"In this note, we show that the COS method can be adapted to make use of the simpler type-2 NUFFT, thus unleashing unprecedented speed for pricing options.","summary":"The COS method is a very efficient way to compute European option prices under Lévy models or affine stochastic volatility models, based on a Fourier Cosine expansion of the density, involving the characteristic function. This note shows how to compute the COS method formula with a non-uniform fast…","strategy":null},"assumptions":{"status":"available","items":[{"assumption":"The asset’s normalized characteristic function is known and can be evaluated.","type":"model","stated":"explicit","quote":"We consider an asset F with a known (normalized) characteristic function"},{"assumption":"The forward price F(0,T) is given by the usual formula for equity forwards.","type":"market","stated":"explicit","quote":"F(0, T) is typically the forward price to maturity T of an underlying asset S. For example, for an equity with spot price S, dividend rate q and interest rate r, we have F(0, T) = S(0)e^{(r−q)T}."},{"assumption":"The Heston model parameters from Le Floc’h (2014) are used as the test case.","type":"model","stated":"explicit","quote":"We use the parameters from Le Floc’h [2014], that is κ = 1, θ = 0.1, σ = 1, v0 = 0.1, ρ = −0.9 and T = 2 years with r = 0%."},{"assumption":"The default relative tolerance of 10⁻⁹ for the NFFT package is sufficient for accuracy.","type":"statistical","stated":"explicit","quote":"used the default relative tolerance (10−9) for the NFFT package."}]},"datasets":{"status":"available","items":[{"name":"Heston model simulation using Le Floc’h (2014) parameters","provider":null,"kind":"synthetic","access":"unknown","period":null,"universe":null,"quote":"We use the parameters from Le Floc’h [2014], that is κ = 1, θ = 0.1, σ = 1, v0 = 0.1, ρ = −0.9 and T = 2 years with r = 0%."}],"known":[],"sample":{"start":null,"end":null,"market":null,"asset_class":null,"frequency":null},"data_resolution":"unclear"},"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":"Using the type‑2 NUFFT yields a very significant speedup for pricing many options of the same maturity, with the threshold around 100 strikes.","kind":"performance","quote":"The COS method can make use of the type-2 NUFFT. This leads to a very significant speedup when many options of the same maturity but different strikes are priced. The threshold is around 100 strikes, and depends on the cost of the characteristic function evaluation."},{"finding":"For M=256 and L=8, the NUFFT implementation prices up to 3 625 k options per second versus 430 k for the classic COS method when 500 strikes are batched.","kind":"performance","quote":"Table 3: Number of options priced per second for a batch of uniform strikes under the Heston model, with L = 8 and various number of points M."},{"finding":"The RMSE and MAE of the NUFFT and classic COS methods are essentially identical (e.g., RMSE 5.62e‑06 for M=256).","kind":"other","quote":"There is a small difference in the root mean square error (RMSE) and the mean absolute error (MAE) between the two methods when M = 1024: we"}],"reported_performance":null,"out_of_sample":{"method":"none","headline_out_of_sample":null}},"independent_evidence":{"status":"none_found","supports":[],"contradicts":[],"qualifies":[],"compared_with":28,"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":1,"extracted_at":"2026-11-14T17:32:12Z"},"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/"}