Trying to understand patterns on a sufficiently deep level for them to be stable, and projecting those patterns forward to arrive at qualitative and rather general predictions not involving e.g. happy fun specific dates, is just what I try to do

Have you had any successes?