The Economist model (link on the first post in this thread) gives Biden (as of today) a 92% chance of winning the electoral college. American polls call it much closer. I wonder why the split? The Economist is a notably even-handed magazine, not given to silliness, but it's also not American. Could distance and the generally anti-Trump attitude of the Brits have something to do with it? Wishful thinking somehow infecting their model? Or is the problem on the American side, bad polling showing its face. 538 rates polls, and some of the get D- (really.) It'd be nice to know where the reality lies.
538 currently has Biden at an 87% chance of winning the electoral college, so not that far from The Economist forecast. 538's past track record is very good. In 2016 they gave Trump a better chance than other poll aggregators at about 30%, which is actually a pretty decent chance of winning. If the weather forecast was a 30% chance of rain, I might put an umbrella in the car.
If you look at the 538 forecast page and scroll down to "The winding path to victory" which shows the closest races on either side of a 270 electoral vote center line, you can see why Trump is currently looking to be in terrible shape to win re-election. Ohio, North Carolina, Arizona, Florida, Pennsylvania, Trump needs to win
all of them; and they are all states in which Biden is currently leading in the poll averages. Of those, Biden only needs to win one, Pennsylvania or perhaps Florida and the deal is done. Florida is really crucial for Trump. If he loses Florida, his path to victory will be
extremely difficult as he would need to perform a repeat of 2016 with Pennsylvania, Wisconsin, and Michigan, which currently looks very unlikely.
https://projects.fivethirtyeight.com/2020-election-forecast/The 538 national polling average is here. Biden is currently in double digits at a +10.5% margin.
https://projects.fivethirtyeight.com/polls/president-general/national/