Artificial Neural Networks for Valuation of Financial Derivatives and Customized Option Embedded Contracts

In this paper we propose and test a valuation methodology for improving the efficiency of contingent claims pricing using Artificial Neural Networks (ANN). Contingent claims is by now a standard method for pricing under uncertainty nonlinear (option embedded) contracts, for both financial options (standardized or customized) and real (investment) opportunities. In the presence of liquid option markets, implied volatility surfaces have aided considerably option pricing with and without the use of ANN. In the absence of such liquid markets, customized positions are much harder to evaluate. The method in this paper improves the efficiency of valuation and financial decision-making dramatically. The method can be used by financial institutions for real time pricing of customized options and investment contracts with guarantees, and valuation under uncertainty of new ventures and the related growth financing instruments. We compare the proposed (hybrid) method with the simple use of ANN for very hard option pricing problems and we demonstrate the method’s superiority. The combination of accurate option pricing and the resulting efficiency are instrumental for the applications we discuss.