Option pricing via Regime Switching models and MultiLayer Perceptrons: a comparative approach

(Journal cited in: MathSciNet, n. MR2031665).

In order to face the determination of option pricing models alternative to the Black and Scholes one, in this paper we propose two approaches. In the first one we take into account a particular parametric stochastic model for volatility: a regime switching one for the dynamics of the logarithmic returns of the underlying. In the second approach we assume that the biases generally observed for the Black and Scholes prices depend on the violation of more than one of the hypotheses of the related model. In particular, as it is often difficult to a priori determine the violated assumptions, we consider a non parametric tool: the multilayer perceptron artificial neural network. Both the approache we developed have been carefully tested on call options written on the UK stock index FTSE-100. The results we obtained show that the considered models allo improvements with respect to the Black and Scholes pricing, that we consider as benchmark.