I don't really get this. It seems like both types of prediction matter quite a bit.

The only way I can interpret it that makes sense to me is something like:

Thinking really hard about the infinity of things that might happen this week is an unproductive way to generate predictions, because the hypothesis space is too large and you're just going to excessively privilege some salient hypothesis.

Is he giving advice about making correct predictions given that you just randomly feel like predicting stuff? Or is he giving advice about how to predict things you actually care about?