Tag
This paper examines whether ML models can beat the random walk benchmark in forecasting USD/CAD exchange rates, finding that only linear regression statistically outperforms the naive model, with SHAP analysis showing short-term lags dominate predictions.