Boundary error control for numerical solution of BSDEs by the convolution-FFT method

Battle Count: 101

The paper is relevant to quantitative trading through its application to option pricing and hedging (delta computation). Accurate BSDE solutions are fundamental for derivatives pricing, risk management, and hedging strategies. However, the paper is primarily a numerical methods contribution rather than a trading strategy paper. The improved boundary error control is particularly relevant for pricing deeply out-of-the-money options, which are common in trading desks. The method could be integrated into pricing engines for exotic derivatives.

The method requires careful implementation of: (1) FFT-based convolution with proper phase-shift handling for centered frequency grids, (2) exponential damping and shifting with periodicity enforcement at each time step, (3) recovery of original functions from damped-shifted representations, (4) proper handling of the Z-component convolution kernel, and (5) stability condition verification (N >= L/(sigma*sqrt(2/Delta_t))). The algorithm is well-specified in Algorithm 1, but numerical stability near boundaries and correct parameter selection (alpha, L, N) require expertise. The backward time-stepping structure adds complexity compared to forward methods.

The paper provides a detailed algorithm (Algorithm 1) with explicit steps for the FFT-based backward iteration, including damping, shifting, and recovery terms. Numerical parameters are specified (S0=100, K=100, r=R=0.01, mu=0.05, sigma=0.2, T=1). However, no code repository is provided, and some implementation details (e.g., exact initialization of Z-component) are left to the practitioner. The error analysis is rigorous with proofs in the appendix.