We use a quantile-boosting approach to compute out-of-sample forecasts of gold returns. The approach accounts for model uncertainty and model instability, and it allows forecasts to be computed under asymmetric loss functions. Different asymmetric loss functions represent different types of...