Classification outperforms regression throughout matched machine studying fashions in portfolio development. A stacking ensemble of gradient boosted bushes, random forest, and neural community yields a value-weighted annualized Sharpe ratio of two.08 for classification...