Are there any Clinton-Trump voters?

Maybe they are here:

"Biden leads among Black voters by an 83% to 8%, or 75-point, margin. That, of course, is a huge advantage for Biden, but it also represents a small improvement for Trump since 2016. Hillary Clinton was ahead of Trump by a 79-point margin among Black registered voters in the pre-election polls taken right before the 2016 election."

Or maybe some Black voters are planning to stay home.

In a timing coincidence, 10 hours after after I posted the question, G. Elliot Morris (of The Economist) e-mailed this out:

CBS News and YouGov released two new polls today, one each in Georgia and North Carolina. The toplines numbers are a bit rosier for Joe Biden than the averages—he leads Donald Trump by 1 and 4 points, respectively—but the more important stuff is under the hood.

One thing that makes polls from YouGov extremely useful is that they asked many of their online panelists who they voted for shortly after the 2016 election. The idea is that asking them shortly after the contest will minimize bias in vote recall; some past research suggested that people are less likely to admit voting for the loser, though that’s no longer clearly the case.

This enables some more-accurate-than-usual comparisons of how people voted in 2016 and how they say they’ll vote now. In the case of the CBS data in Georgia and North Carolina, we get some stunning data on how likely Trump and Clinton’s 2016 voters are to vote for the same party’s candidates this time around. They find that 2% of Hillary Clinton’s 2016 voters say they’re going to vote for Donald Trump this November, whereas 6% of Trump’s 2016 voters say they’re going to cast ballots for Joe Biden.

(Wonky note: Although these differences are technically within the normal margin of sampling error, the interval is smaller when percentages are closer to 0—making the differences more likely to be accurate—and also match a lot of other polls.)