Machine learning portfolio allocation

We find economically and statistically significant gains when using machine learning for portfolio allocation between the market index risk-free asset. Optimal rules time-varying expected returns volatility are implemented with two Random Forest models. One model is employed in forecasting monthly excess macroeconomic factors including payout yields. The second used to estimate prevailing volatility. Reward-risk timing provides substantial improvements over buy-and-hold utility, risk-adjusted returns, maximum drawdowns. This paper presents a unifying framework applied both return- volatility-timing.