ABSTRACTWe introduce a novel framework that dynamically optimizes currency factor strategies via trading currency spot and forward. We examine the performance of 24,336 portfolio optimization approaches and find that the optimized currency factors significantly outperform the naïve factors after correcting for data snooping bias. Our framework suits both symmetric factor portfolios, including carry, momentum, and value, and asymmetric factor portfolios, such as time series momentum and return signal momentum. An out‐of‐sample procedure that aggregates all the outperforming optimization approaches validates the economic significance of our optimized factor portfolio.