A binomial model for pricing US-style average options with reset features

We develop a pricing algorithm for US-style period-average reset options written on an underlying asset which evolves in a Cox-Ross Rubinstein(CRR) framework. The averaging feature of such an option on the reset period makes the price valuation problem computationally unfeasible because the arithmetic average is not recombining on a CRR tree. To overcome this obstacle, we associate to each node of the lattice belonging to the reset period a set of representative averages chosen among all the effective arithmetic averages attained at that node. On the remaining time to maturity, a US period-average reset option becomes a US standard one and the Barone Adesi-Whaley approximation is used to compute an option value in correspondence to each representative average lain at the end of the reset period.