Pricing American options: RNMs-constrained entropic least-squares approach

Recently the entropy-based valuation of European options (Stutzer, 1996) has been extended to American option pricing. In this paper, we improve the pricing accuracy by incorporating informative risk-neutral moments (RNMs), which are recovered from a set of market-available option data, as constraints into the entropy framework. With these RNMs, an appropriate risk-neutral measure close enough to the correct one is achieved. An adjusted least-squares algorithm is then utilized to determine the optimal exercising strategy. The results based on simulations and empirical analysis demonstrate that our method can price American options rather accurately and significantly outperforms the benchmark methods.