To cite this version:Rama Cont, Peter Tankov Abstract. We propose a stable nonparametric method for constructing an option pricing model of exponential Lévy type, consistent with a given data set of option prices. After demonstrating the ill-posedness of the usual and least squares version of this inverse problem, we suggest to regularize the calibration problem by reformulating it as the problem of finding an exponential Lévy model that minimizes the sum of the pricing error and the relative entropy with respect to a prior exponential Lévy model. We prove the existence of solutions for the regularized problem and show that it yields solutions which are continuous with respect to the data, stable with respect to the choice of prior and converge to the minimum-entropy least square solution of the initial problem.Key words. inverse problem, entropy, Lévy process, model calibration, option pricing, regularization.
AMS subject classifications. 49N45 60G51 60J75 91B701. Introduction. The specification of an arbitrage-free option pricing model on a time horizon T ∞ involves the choice of a risk-neutral measure [24]: a probability measure Q on the set Ω of possible trajectories (S t ) t∈[0,T∞] of the underlying asset such that the discounted asset price e −rt S t is a martingale (where r is the discount rate). Such a probability measure Q then specifies a pricing rule which attributes to an option with terminal payoff H T at T the value C(For example, the value under the pricing rule Q of a call option with strike K and maturity T is given byGiven that data sets of option prices have become increasingly available, a common approach for selecting the pricing model Q is to choose, given option prices (C(H i )) i∈I with maturities T i payoffs H i , a risk-neutral measure Q compatible with the observed market prices, i.e. such that C(. This inverse problem of determining a pricing model Q verifying these constraints is known as the "model calibration" problem. The number of observed options can be large ( 100 − 200 for index options) and the Black-Scholes model has to be replaced with models with richer structure such as nonlinear diffusion models [17] or models with jumps [12]. The inverse problem is ill-posed in these settings [13,33] and various methods have been proposed for solving it in a stable manner, mostly in the framework of diffusion models [3,4,5,7,14,17,25,32,33].We study in this paper the calibration problem for the class of option pricing models with jumps -exponential Lévy models-where the risk-neutral dynamics of the logarithm of the stock price is given by a Lévy process. The problem is then to choose the Lévy process -described by its Lévy measure-in a way compatible with a set of observed option prices. Option prices being evaluated as expectations, this inverse problem can also be interpreted as a (generalized) moment problem for a Lévy process: given a finite number of option prices, it is typically an ill-posed problem. The relation between the option prices and the Lévy measure being nonlinear, we ...