Results vs polls 40 days out
| Election | Polls 40 days out | Actual | Change |
| 1976 | 4.9% | 2.1% | -2.8% |
| 1980 | 4.6% | 9.7% | 5.1% |
| 1984 | 16.9% | 18.2% | 1.3% |
| 1988 | 5.1% | 7.8% | 2.7% |
| 1992 | 11.4% | 5.6% | -5.8% |
| 1996 | 13.9% | 8.5% | -5.4% |
| 2000 | 0.6% | -0.5% | -1.1% |
| 2004 | 5.0% | 2.4% | -2.6% |
| 2008 | 1.1% | 7.2% | 6.1% |
| 2012 | 3.8% | 3.9% | 0.1% |
| 2016 | -2.8% | -2.1% | 0.7% |
| 2020 | 7.4% |
I’ve compiled this table showing the average of the polls 40 days before a US presidential election (from 538) and the actual result in terms of popular vote.
In 2012 and 2016 the popular vote didn’t shift that much in the last 40 days.
In 2008 Obama went from a narrow lead to a large lead.
In 2000 and 2004 Bush’s lead shrank in the last 40 days.
Clinton in both elections had a huge lead 40 days out, but halved by the election.
What is interesting in 2020 is how little movement there is. The smallest the gap has been since May is 6.1% and the largest is 9.6%.
But if anything can give Trump a second term, the Democrats promising to stack the Supreme Court by adding on additional Justices could do it.