Abstract
Pricing American options accurately is of great theoretical and practical importance. We propose using machine learning methods, including support vector regression and classification and regression trees. These more advanced techniques extend the traditional Longstaff-Schwartz approach, replacing the OLS regression step in the Monte Carlo simulation. We apply our approach to both simulated data and market data from the S&P 500 Index option market in 2019. Our results suggest that support vector regression can be an alternative to the existing OLS-based pricing method, requiring fewer simulations and reducing the vulnerability to misspecification of basis functions.
Highlights
Pricing American options has been a challenging problem in the financial industry from the perspective of both academic research and practical applications
Considering that some machine learning methods may outperform in small datasets and capture the complicated relationship between explanatory variables, we propose two alternatives: support vector regression (SVR) and classification and regression trees (CART)
We consider several machine-learning alternatives to the classic linear regression employed in the Longstaff-Schwartz algorithm, including support vector regression (SVR) and classification and regression tree (CART)
Summary
Pricing American options has been a challenging problem in the financial industry from the perspective of both academic research and practical applications. The idea of combining Monte Carlo methods with approximations of the continuation value in backward induction schemes goes back to Carriere (1996) He estimates the continuation value function at each possible exercise date by minimizing squared errors using smooth splines and local regressions, and compares the immediate payoff of exercising options with the continuation value, to determine the exercise policy. Empirical results show that SVR outperforms other procedures in the simulated data environment in terms of root mean squared error (RMSE) This finding is robust across different moneyness, which showcases that our method can be a good alternative to traditional OLS-based option pricing methods. When it comes to real market data, machine learning techniques do not continue to outperform others.
Sign-in/Register to view highlights & summary Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
