In case you missed this post yesterday, I want to flag again how freaky-deaky sharp this poll has been — for a long time — in predicting the national popular vote with its final estimate. The only meaningful miss was in 2000, and the last days before Election Day that year were chaotic because of the last-minute oppo drop about Bush’s DWI arrest years before.

What did the @ABC-WP tracking poll look like 1 week out in past elections? Here you go: pic.twitter.com/HKMm5AksDS

— Karen Travers (@karentravers) November 1, 2016

ABC/WaPo just released its final estimate this year. Result: 47/43 in the four-way race, 49/46 in the two-way. The reason it tightens up slightly in the two-way is because there are still more Republicans (eight percent) than Democrats (five percent) who are voting third-party as of right now. A few highlights from this morning’s tracking poll, the findings of which weren’t materially different from this afternoon’s:

While the contest is a close one, Clinton has held a numerical advantage almost continuously during the campaign. The average result among likely voters in ABC/Post polls since June is 47 percent for Clinton, 42 Trump, 5 Johnson and 2 Stein — almost identical to the numbers today…

Thirty-six percent of likely voters in the latest (in this case, single day) results say they’ve already voted, favoring Clinton by 53 to 43 percent, suggesting greater success for her ground game, at least in this respect…

The education gap is especially wide among whites. Among those with a college degree, it’s 47 percent Clinton to 41 Trump. Among whites without a college degree — Trump’s mainstay, it’s 62 percent Trump to 29 Clinton. On this result, as in others, what matters ultimately is how many in each group — whether with positive motivations or negative ones — get out and vote.

So in the end, if you believe ABC/WaPo, Hillary’s year-long lead among white college grads will stand up tomorrow despite the fact that Republicans have been coming home to Trump over the last few weeks. No Democrat has won that group in decades; Romney took them by six points, so Hillary’s on the verge of a 12-point swing here from 2012. If she wins the election narrowly, they’re likely to be a key reason why. The other major boost she’s getting comes from Latinos, whom ABC/WaPo sees breaking 71/19 for her. Obama famously won Latinos 71/27 in 2012, which some thought represented rock bottom for a Republican. Some polls during this campaign, in fact, showed Trump performing comparably to Romney with Latinos despite his border-hawkery. Not this one. In the last few days, in fact, polls of Latinos have turned grim for him: A new one from New Latino Voice has him trailing 71/23, which is slightly off Romney’s pace, and Latino Decisions has it 79/18, which is way off the pace. If Trump loses white college grads and gets swamped by massive Latino support for Clinton, his chances become a pure function of turning out working-class whites in gigantic numbers. He is outpacing Romney with that group in the ABC poll, winning them by 33 points. This is why the Rust Belt has come back on the board for him. His base of whites without a degree are going to put a scare into Clinton there.

By the way, since I know you’re wondering: Is there any chance that Trump loses the popular vote, as ABC/WaPo suggests he will, but wins the Electoral College? Why yes, actually. A week ago, when Clinton had a lead of 4.7 points in the national average, Nate Silver gave that a considerable 10 percent chance of happening. The reason is that Clinton’s support isn’t evenly distributed. She’s doing better than Obama did four years ago in red states where Trump is going to win anyway but not as well as Obama did in battlegrounds. It’s possible that she piles up votes in blue strongholds and red strongholds (in a losing effort) and then loses narrowly in multiple swing states like Florida and North Carolina. Clinton’s national lead has shrunk since Silver wrote that post so presumably the odds of Trump winning the EC while losing the popular vote have risen further since then. Having a Gore/Bush-style split running either way in this election would probably be the worst possible outcome for an already divided nation, so expect it to happen.