Restricted Entropic and Quadratic Hedging in the Small-Risk-Aversion Limit

Abstract

In incomplete markets, contingent claims cannot in general be replicated exactly, and hedging must therefore be formulated as an optimization problem over admissible trading strategies. This paper studies the relation between quadratic hedging and entropic hedging in a discrete-time incomplete market when strategies are restricted to bounded finite-dimensional parametric classes. These restrictions induce admissible gain sets that are compact and convex in \(L^2\), providing a common well-posed setting for both the restricted mean-square hedging problem and the restricted exponential-utility hedging problem. The main contribution is to establish a uniform small-risk-aversion expansion of the entropic risk functional over the restricted gain set and to use this uniformity to obtain convergence of the corresponding minimizers. Under a uniform exponential integrability assumption, the restricted entropic hedging gains converge in \(L^2\) to the unique restricted quadratic hedge as the risk-aversion parameter tends to zero. The theoretical result is illustrated numerically in a regularized discrete-time stochastic volatility model of Heston type.