Foreign exchange trading and management with the stochastic dual dynamic programming method

Abstract We present a novel tool for generating speculative and hedging foreign exchange (FX) trading policies. Our solution provides schedule that determines trades in each rebalancing period based on future currency prices, net account positions, incoming (outgoing) flows from business operations. To obtain such policies, we construct multistage stochastic programming (MSP) model solve it using the dual dynamic (SDDP) numerical method, which specializes solving high-dimensional MSP models. our methodology within an open-source SDDP package, avoiding implementing method scratch. measure performance of FX prices as mean-reverting process with random events incorporate trends. calibrate this price seven pairs, demonstrating policies not only outperform benchmarks currency, but may also be close to ex-post optimal solutions. show how can used generate more or less conservative strategies by adjusting risk tolerance, variety contexts time scales, ranging intraday monthly Finally, examine impact increasing trade policy uncertainty (TPU) levels findings. findings volatility currencies emerging economies rises comparison developed markets. discover increase TPU level has no effect average profit obtained method. However, exposure increases (decreases) group (developed)