AiChih Chang¹, Jim (Junmin) Shi²

(1. Rutgers University, USA; 2. New Jersey Institute of Technology, University Heights, USA)

 

 

Abstract: Pricing American options is notoriously intractable or computationally challenging due to its complexity feature of dynamic exercise strategy and the inherent uncertainty pertaining to its underlying asset price. Numerically, this study leverages the Least Squares method to price American options based on two simulation methods. In particular, to simulate the price process of the underlying asset, we propose the Transform-Expand-Sample (TES) approach, and compare its performance with the benchmark model of random walk. Random walk method is widely used if the volatility of the underlying asset price is the only factor affecting its behavior. In contrast, the TES approach is a versatile methodology for modeling stationary time series, whose principal merit is its ability to simultaneously capture first-order (marginal distribution) and second-order (autocorrelations) statistics of empirical time series. We experiment with several real-market American call options to illustrate the implementation of those two models. With an acceptable accuracy, the estimated option prices obtained by both approaches match the actual market price of the American option.

 

 

Key words: American option pricing; simulation; TES

 

JEL code: C630