When it comes to (forecasting and) combining, the recipe is simple: when in doubt… combine!

But that comes at a cost, as in within that combination there are good and bad forecasts., forecasts coming from better and worse forecasters (if the latter are judgmental forecasts).

So, if you knew who will give you the best forecasts in advance, you would not have to combine in the first place; but you don’t, and thus the doubt!

Therefore, in the absence of any (reliable) information of who will consistently provide better forecasts, stick to (usually equally weighted) combining…