Like a star-crossed baseball team trying to close out a pennant, Mrs Hillary Clinton holds an advantage in the race for president that feels real yet not at all reassuring - for her partisans or for Trump-fearers worldwide.
"She just has to hang on" seems like the dominant emotion among her backers and, for that matter, among her staff, who handled her ground-zero meltdown like a front office desperately trying to minimise the seriousness of a star pitcher's injury: "Just a bruise... OK, a muscle pull... we took him out just as a precautionary measure... he'll make his next start... OK, fine, he's on the disabled list."
Mrs Clinton is still a good bet to hang on but, last month, amid the Khan affair, it seemed as if Mr Donald Trump might simply plunge permanently into McGovern- Goldwater territory, leaving the actual election a mere formality.
Instead, he has stayed alive, closing back to within a few points in the polling average. His odds are still poor - in the most plausible scenarios, he loses - but he still has a path, and here is what he needs to walk it.
First, Mr Trump wants this to stay a four-way race. In national polling, his ceiling is close to Mrs Hillary's floor - he peaks in the low 40s, she peaks close to 50 per cent. This suggests that there are more true #NeverTrump than #NeverHillary voters, and that a section of voters (especially millennials) have rejected him pre-emptively but are still considering non-Clinton options. In which case, candidates Jill Stein and Gary Johnson are likely to take more votes from Mrs Clinton than from Mr Trump.
If they drop - as third-party candidates tend to do - towards statistical insignificance, then Mr Trump will need to win over a lot of undecideds who plainly do not want to vote for him. But this is a weird year, and they may not drop. If Mr Johnson can equal Mr John Anderson in 1980 and Dr Stein can equal Mr Ralph Nader's fateful 2000 run (and they are polling at roughly those levels today), then you can imagine a final line like Trump 44, Clinton 43.5, Johnson 7, Stein 3 - the stuff of Democratic nightmares but, for the Trump campaign, a still possible dream.
Second, Mr Trump needs to turn out a lot of people (working-class whites, in particular) who rarely or almost never vote. In the primary campaign, he did not usually outperform his polls, and there was little evidence of "Shy Trumpers" telling pollsters they were for Senators Marco Rubio or Ted Cruz and then pulling the lever for Mr Trump in secret.
But pollsters can never be certain about the composition of the electorate, and a November election throws up fewer obstacles to the casual voter than a primary campaign (no caucusing, no party-registration requirements, everybody knows when Election Day is, and so on). Past celebrity candidates like Jesse Ventura and Arnold Schwarzenegger probably benefited from disaffected, usually apolitical voters coming out for them, and there is no reason to assume it could not happen to some extent for Mr Trump.
Here, it is noteworthy that the best poll for Mr Trump in this cycle, the USC Dornsife/Los Angeles Times daily tracking poll, includes a larger sample of people who did not cast a ballot in 2012 but plan to vote in 2016 - and Mr Trump is cleaning up with them. He is not cleaning up in the poll as a whole but he does often lead it and, when he does, his leads are in roughly the same territory as the four-way scenario just sketched: Trump 44, Clinton 43; Trump 45, Clinton 42.
This points to the third key to a Trump victory: He needs Mrs Clinton's potential support to be depressed, which means he needs outside events to raise constant doubts about her leadership and, beyond that, about the entire political establishment that she embodies. Earlier this year, I called Mr Trump's ideal event a "grey swan" as opposed to a black one - meaning that he wants a constant drippage of stuff that makes both Mrs Clinton and the larger elite seem clueless, feckless or corrupt, but not the kind of crisis that would change the shape of the campaign or make him an untenable risk.
Weeks before, the Republican convention was filled with grey swans - terror attacks in Europe, cop killings here at home, Federal Bureau of Investigation director James Comey's rebuke of the Democratic nominee - and, not surprisingly, they were weeks when Mr Trump sometimes led the polls. There have been fewer such events since, but Mrs Clinton's hidden-then-acknowledged pneumonia is a perfect example of what Mr Trump needs. It is not a black swan; not a devastating illness that would force the Democrats to turn to the more electable Mr Joe Biden. Instead, it just feeds into two of Mr Trump's narratives: alpha-male power versus actual physical weakness, and bold outsider truth-telling versus reflexive elite cover-up.
We will see how the polls look when the pneumonia incident is baked in. But I would not be shocked to see a few tied at, say, 43-43.
Now, let me turn the screw a little further. The Electoral College is an unusual system, and Mr Trump is an unusual candidate. He is likely to underperform among normal Republicans in many red states, where the white working class is already very Republican, by losing white suburban professionals who voted for Senator John McCain and former governor Mitt Romney. But he might overperform in Rust Belt states where the white working class is still a residually liberal swing vote, and where there are a lot of disaffected independents who sat out 2012. (That is probably how you can have state polls showing strikingly close races in Republican strongholds like Georgia and Arizona, even though Mr Trump is quite competitive in swing states like Ohio.) This unusual combination - underperforming but still probably winning Republican states, possibly overperforming in purple states - suggests a true black swan endgame: Not Trump 44, Clinton 43, but Clinton 45, Trump 43 - except that Mr Trump, with his Rust Belt strength, loses a lot of reliable deep-red votes he does not need and turns out just enough non-voters in a few key swing states to take the Electoral College, 270-268.
No, it is not likely. No, do not freak out. But for this race to end with a huge Electoral College crisis is the kind of outcome everything that has happened this year almost - almost - leads one to expect.
A version of this article appeared in the print edition of The Straits Times on September 15, 2016, with the headline 'How Trump may win'. Print Edition | Subscribe
We have been experiencing some problems with subscriber log-ins and apologise for the inconvenience caused. Until we resolve the issues, subscribers need not log in to access ST Digital articles. But a log-in is still required for our PDFs.