Option Pricing Using Machine Learning Models: A Comparative Study with the Black-Scholes Model

Abstract

The capital market has always been accompanied by significant fluctuations, influenced by a variety of factors. These conditions expose market participants to different types of risks, thereby increasing the necessity of employing risk management methods. One of the widely used instruments in this context is the option contract. However, one of the main challenges related to options is their pricing. Among the most well-known models for option pricing is the Black-Scholes model, which is extensively used in practice. Nevertheless, due to its simplifying assumptions, its outcomes do not always align with real market conditions. Machine learning models, owing to their high flexibility and ability to analyze complex datasets, hold strong potential for improving the accuracy of option pricing. This study focuses on pricing options using machine learning algorithms. In this research, 181 option contracts traded on the Tehran Stock Exchange from April 2018 to October 2024 were examined. After data collection and standardization, the option prices were predicted using both machine learning algorithms and the Black-Scholes model. The predicted prices were then compared with actual market prices using Mean Absolute Error and Root Mean Squared Error as evaluation metrics. Additionally, the models’ performance was analyzed based on contract maturity and moneyness. The results indicate that machine learning algorithms outperformed the Black-Scholes model both overall and within the maturity and moneyness subcategories.