Efficient Variations of the Fourier Transform in Applications to Option Pricing

In this paper, we clarify the relationships among popular methods for pricing European options based on the Fourier expansion of the payoff function (iFT method) and the simlified trapezoid rule. We suggest new variations that allow us to decrease the number of terms by a factor of between five and ten (when the iFT requires several dozen terms), or even by a factor of several dozen or a hundred (when the iFT may need thousands or millions of terms). We also give efficient recommendations for an (approximately) optimal choice of parameters for each numerical scheme.