A Markov Chain Approximation for American Option Pricing in Tempered Stable-GARCH Models

This paper considers the American option pricing problem under the stochastic volatility models. In particular, we introduce the GARCHmodel with two heavy-tailed distributions: classical tempered stable (CTS) and normal tempered stable (NTS) distribution. Then we apply the Markov chain approach to compute the prices of American style options under these two models. Minimal entropy provides a convenient way to construct equivalent martingale measure (EMM) and allows us to overcome the difficulties in incorporating the Markov chain approximation. The convergence of the approximation is also proved. Both numerical and empirical results are analyzed to show the advantages and drawbacks of our approach.