The election is two weeks from today. When I took an into psych course over half a century ago, I was taught in it that two weeks is the period of short term memory, the period in which we remember events with special salience. I do not know if this is still the official view of the profession, but it has since then made sense to me: I seem to be able, even now, to remember what happened day by day for the previous two weeks. Things before then are "in the past," although certainly some are salient and on my mind. But those that happened in the past two weeks are just that much more on my mind.
With this in mind even four years ago when people asked me to forecast the election outcome I would drag this up and say "anything can happen in the last two weeks that can change it," and four years ago it happened with the James Comey public reopening of an email investigation into Hillary Clinton 11 days before the election. Even though about two days before he announced nothing was found, the damage was done. This year we all remember this, and while he is further ahead in national polls than she was at this point then, Joe Biden is not much further ahead, and even behind in some, than she was in those crucial battleground states that will determine the outcome. So it remains fully possible that something unexpected can happen that will give Trump the victory.
I must admit, however, that I have been trying to think what could do it. Much discussion focuses on "October Surprise," as if things early in October has as much salience as those in the last two weeks. So far most of those surprises have hurt Trump more than Biden, and the poll gap has widened in Biden's favor, with the new rise in cases and hospitalizations of the coronavirus seeming to be the dominant issue, and with Trump's illness and superspreader events not helping him on that front. So if something happens to push it the other way, it is going to have to overcome a strong pressure coming on that front that I simply do not see moving in Trump's favor. There will not be a vaccine approved prior to Nov. 3, much to Trump's distress and despite all his efforts to force one through.
Of course we have seen the Trump people try to push new stuff on the Hunter Biden case and Burisma, with last Wednesday's New York Post story about his supposed laptop. But, not only did it come out before the final two weeks, it does not seem to be convincing anybody not already in the Trump camp, just too many holes and nonsense I shall not bother with. Yes, the brief blocking of it by Twitter et al gives it a few more legs, but it seems not to be going anywhere serious, mostly just another effort to get the Trump base out, if it was not already out.
My suspicion is that at this point the only thing that could really do it would be a genuinely unexpected event, with something from abroad the most likely, given that we pretty much know all there is to know about Joe Biden himself. A 9/11 style terror attack from the Middle East might do it, fits Trump's narrative and would allow him to pull a "rally the troops around the national leader," or something equivalent.
I close by noting a report I just saw that looked at October Surprises, not just final two weeks stuff, over elections dating back to 1980. In fact most of them were pretty well baked in by the beginning of October, with few seeing movement in national polls exceeding 1%. 2016 saw the second largest move, almost 3%. That was the Comey Announcement Effect. Curiously the only other election with a larger net move was 1992, when Clinton gained about 7%, although I do not remember a specific event or "surprise" that triggered that. So, in fact, the probability of a something really election altering happening within the next two weeks is pretty low, although not yet to be ruled out.