A quantile-boosting approach to forecasting gold returns

• We apply a quantile-boosting approach to forecast gold returns.
• The approach accounts for model uncertainty and model instability.
• We use the approach to study the economic value-added of forecasts.
• We compare the approach with the lasso quantile-regression approach.

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 investors (optimists versus pessimists). We document how the performance of a simple trading rule varies across investor types.