The quality of an exchange rate forecasting model has typically been judged relative to a random-walk in terms of out-of-sample forecast errors. The difficulty of outperforming this benchmark is well documented, although Clarida and Taylor have demonstrated how the random walk can be beaten in...