A bi-weekly challenge from Andre Mirabelli & Opturo (August 31, 2020)

How does an attribution modeler address the following challenge?
Karnosky and Singer provided an ex post decision attribution methodology that attributed active returns to currency decisions in an investment process that could actually be carried out. Are there ever valid reasons, in an ex post decision attribution process, for attributing active returns to currency-related properties that can never be implemented as steps in an actual investment process?