Option pricing, Black-Scholes Model, Monte carlo simulation, Neural networks

Abstract

           This study aims to compare option pricing calculations employing a numerical method and a neural network for European options, utilizing daily data from a sample group of NEX-W2 warrants officially traded on the Stock Exchange of Thailand. NEX-W2 warrants have a life span of three years from issuance and can be exercised only once upon expiration.

           From the study, it was found that the option prices calculated through the numerical method, consisting of the Black-Scholes method and the Monte Carlo method, were very close to each other. However, the option prices derived from the neural network method were higher than those calculated using the numerical method and were closely comparable to the actual market option prices. Notably, the price movement direction of each method was found to correspond with the market price movement direction.

          Considering the RMSE value of the three methods, it was discovered that the neural network model yielded the lowest RMSE value of 1.02. This suggests that the neural network model is most appropriate for approximating option contract prices in this case study. Meanwhile, the RMSE values for the Black-Scholes method and the Monte Carlo method were 3.936849 and 3.937826 respectively.