Just for the funzies, let's look at how elections turned out since the last great realignment, compared to where they stood about now:
Presidential national average lead, 17 days before E-day:
1992: Clinton +11.1*
1996: Clinton +13.8*
2000: Bush +3.1*
2004: Bush +3.4
2008: Obama +5.6
2012: Obama +0.1
2016: Clinton +6.4
2020: Biden: +10.6
* for the month of October as a whole.
Final stats, popular vote:
1992: Clinton +5.6 (5.5 toward Rep)
1996: Clinton +8.5 (5.3 toward Rep)
2000: Gore +0.5 (3.6 toward Dem)
2004: Bush: +2.5 (0.9 toward Dem)
2008: Obama: +7.3 (1.7 toward Dem)
2012: Obama: +3.9 (3.8 toward Dem)
2016: Clinton: +2.1 (4.3 toward Rep)
Average shift: +0.85 toward Rep
Predicted Biden margin: +5.1 to +14.4 (median +9.75)
Predicted EVs, assuming the same partisan lean as 2016: 308-230 to 422-125
Conclusions: October polls are a strong indicator of the end result and either a) the Clintons always underperform their polls badly; or b) stronger third-party voting correlates with Republicans making up some ground.