Update 5 Nov: It looks like my prediction of a close election was largely accurate, and more accurate than most media outlets and polls.
My final gut instinct prediction for the 2020 US Presidential election is: too close to call. I give Biden a 55% chance of victory and Trump a 45% chance. This is largely based on apparent late gains by Trump making many swing states too close to call. I give Biden a slightly better chance based on his small polling lead in Pennsylvania.
Also, I understand that my preference for Biden over Trump could be biasing the estimate. I have a general tendency of being pessimistic about the side I support when I think that the contest is close (but not when I don’t expect it to be close, e.g. US 2016 or Australia 2019, in both of which I was falsely confident).
For the record, here is my past predictions track record:
Ones that I got right:
UK 2017: May 99%; Corbyn 1%; May won but she only just got over the line
Canada 2019: Trudeau 60%; Scheer 40%; Trudeau won but lost the popular vote
UK 2019: Johnson 99%; Corbyn 1%; Johnson won in a landslide
Dem Primaries: Biden 90%; Sanders 5%; Buttigieg 5%; Biden won despite early losses
I also correctly predicted the result of the gay marriage referendum in Ireland and Australia (both yes).
Ones that I got wrong:
US 2016: Clinton 85%; Trump 15%; Trump won (my worst prediction ever)
Australia 2019: Shorten 70%; Morrison 30%; Morrison won
I also wrongly predicted the result of the Brexit vote (I gave Remain 2/3 chance).