This paper provides a comprehensive analysis of machine learning methods applied in short-term exchange rate forecast and portfolio evaluation. The models we used include: random forest, gradient boosting decision tree (GBDT) and extreme gradient boosting (XGBoost). We provide evidence that...