Could I have predicted Trump would lose the post-election? Is there a consistent, reliable principle? There is.
The fact I couldn't tell you what he was thinking would have allowed me to safely conclude he was quite fucked.
What I'm thinking is generally competent. If I don't know what you're thinking, it's going to be far from competence, and thus will fail, barring miraculous intervention.
Looking for instances where this principle does not retrodict effectively, I come up with The Last Psychiatrist and Spice and Wolf, which have an unquestionably different feel. In each case I can tell there's something above my understanding. Rather than baffled, I'm curious. Trump is quite intelligent (principle: journalists are the opposite of true) but naturally he's no match for a properly trained scholar. He's thoroughly incapable of thinking anything beyond my capacity.
It didn't occur to Trump to secure the election process, even though he was obviously the victim of severe fraud in 2016. (The real vote was something like 40-58 for Trump.) It likewise did not occur to him to learn from his failures in office, and thus he continued to fail his way out of office. Or, for that matter, to learn from his successes and repeat them. E.g. his black vote skyrocketed because it turns out ninjas prefer to have jobs over not having jobs. However, Trump did not have the necessary training to trace this successful job-unblocking, or others like it, which meant he could not double down. Don't be like Trump: get this training in ASAP.