Hybrid method based on neural networks and Monte Carlo simulation in view of a tradeoff between accuracy and computational time

Abstract

The aim of this article is to introduce a hybrid method, based on the combination of Monte Carlo simulation and neural networks, which ensures for a general model an optimal compromise between accuracy and computing time. The major contribution of this work is that the aforesaid improvements are made whatever the hypotheses adopted to simplify the reality of the studied problem. In this article, this methodology is applied to the option pricing as a problematic in the fields of finance. Based on a database of 6 contracts of European calls on the CAC 40 index, the obtained results are in accordance with our expectations. Indeed, they highlight the precision of this methodology and the decrease of the computational time, in comparison with that of the Monte Carlo simulation. It, thus, implied that the proposed method can be used as an alternative solution to any derivative pricing problem.

Keywords:

• Neural networks
• Monte Carlo simulation
• hybrid method
• computational time
• accuracy
• option pricing

Notes

1 This section is adapted from Jerbi (2006, 2016).