Dirani, Mattia
(2026)
Computing Greeks under CGMY Model with Full Error Control.
[Laurea magistrale], Università di Bologna, Corso di Studio in
Greening energy market and finance [LM-DM270], Documento full-text non disponibile
Il full-text non è disponibile per scelta dell'autore.
(
Contatta l'autore)
Abstract
Fourier pricing techniques are known to fail when computing Greeks under many popular Lévy models, since the interchange of differentiation and integration required by these methods breaks down whenever the characteristic function of the log-asset price decays too slowly. This issue is even more severe for digital options, whose discontinuous payoff further hinders standard approaches. This thesis proposes an alternative methodology that circumvents this problem and delivers arbitrarily accurate estimates of Delta and Gamma under Lévy models admitting a representation as subordinated Brownian motions, such as the CGMY model. Conditionally on the subordinator, the log-price is normally distributed, so that European option prices and Greeks admit closed-form Black-Scholes expressions. Since the subordinator of the CGMY process cannot be simulated exactly, its cumulative distribution function is approximated by numerically inverting its characteristic function via the Fourier-cosine (COS) method. Random samples are then generated using the inverse transform method, and Greeks are obtained by Monte Carlo averaging over the simulated subordinator values, thereby avoiding the differentiation-integration interchange that causes Fourier techniques to fail. A key advantage of the proposed Conditional Black-Scholes (CBS) approach is its explicit error control: the practitioner specifies a probability and an error tolerance, and the algorithm guarantees that the resulting estimate lies within that tolerance of the true value with the requested probability. An explicit error analysis links the COS approximation tolerance, the number of Monte Carlo simulations, and the resulting accuracy. Numerical experiments under the CGMY model confirm that the CBS approach remains accurate where classical Fourier techniques completely fail, and outperforms established benchmarks in terms of both accuracy and computational efficiency.
Abstract
Fourier pricing techniques are known to fail when computing Greeks under many popular Lévy models, since the interchange of differentiation and integration required by these methods breaks down whenever the characteristic function of the log-asset price decays too slowly. This issue is even more severe for digital options, whose discontinuous payoff further hinders standard approaches. This thesis proposes an alternative methodology that circumvents this problem and delivers arbitrarily accurate estimates of Delta and Gamma under Lévy models admitting a representation as subordinated Brownian motions, such as the CGMY model. Conditionally on the subordinator, the log-price is normally distributed, so that European option prices and Greeks admit closed-form Black-Scholes expressions. Since the subordinator of the CGMY process cannot be simulated exactly, its cumulative distribution function is approximated by numerically inverting its characteristic function via the Fourier-cosine (COS) method. Random samples are then generated using the inverse transform method, and Greeks are obtained by Monte Carlo averaging over the simulated subordinator values, thereby avoiding the differentiation-integration interchange that causes Fourier techniques to fail. A key advantage of the proposed Conditional Black-Scholes (CBS) approach is its explicit error control: the practitioner specifies a probability and an error tolerance, and the algorithm guarantees that the resulting estimate lies within that tolerance of the true value with the requested probability. An explicit error analysis links the COS approximation tolerance, the number of Monte Carlo simulations, and the resulting accuracy. Numerical experiments under the CGMY model confirm that the CBS approach remains accurate where classical Fourier techniques completely fail, and outperforms established benchmarks in terms of both accuracy and computational efficiency.
Tipologia del documento
Tesi di laurea
(Laurea magistrale)
Autore della tesi
Dirani, Mattia
Relatore della tesi
Correlatore della tesi
Scuola
Corso di studio
Indirizzo
ENVIRONMENTAL FINANCE
Ordinamento Cds
DM270
Parole chiave
Error analysis,Greeks,Lévy models,Monte Carlo simulation,Option pricing,CGMY model,Variance reduction,Fourier methods
Data di discussione della Tesi
28 Ottobre 2026
URI
Altri metadati
Tipologia del documento
Tesi di laurea
(NON SPECIFICATO)
Autore della tesi
Dirani, Mattia
Relatore della tesi
Correlatore della tesi
Scuola
Corso di studio
Indirizzo
ENVIRONMENTAL FINANCE
Ordinamento Cds
DM270
Parole chiave
Error analysis,Greeks,Lévy models,Monte Carlo simulation,Option pricing,CGMY model,Variance reduction,Fourier methods
Data di discussione della Tesi
28 Ottobre 2026
URI
Gestione del documento: