An Anatomy of Generalized Entropic Estimators in Option Pricing

We adopt a family of nonparametric Cressie-Read estimators to price options based on relative pricing using the underlying asset returns. We use option models with stochastic volatility and jumps to investigate the ability of each member in this family to price options with different moneynesses and maturities. We show that their ability to price options crucially depends on the magnitude of the risk premium implicit in each source of risk and that higher moments of underlying returns are essential to identify large levels of risk premia. Price intervals given by an economically motivated range of Cressie-Read risk-neutral measures usually contain the true option prices, suggesting that our method can be used to price options in illiquid markets. An empirical application with a large cross-section of S\&P 500 options documents the pricing ability with real data and confirms that risk premia is a fundamental determinant of the performance of these estimators