Who will win today's US presidential election?

The answer is, of course, that I don't have a scooby. If the statistical experts who devote their lives to studying this question say the race is as hard to read as a coin toss (and they do), nobody else has got much of a prayer. And I say that with all due respect to the blessed Rory Stewart, who is claiming certain knowledge that Kamala Harris will win by miles, probably because he thinks it will help his punditry career if he's guessed correctly, and that everyone will swiftly forget he ever made a prediction otherwise.  We won't, Rory.  The people of "the Middleland" never forget.

I do have a few observations, though.  The betting exchanges currently say Donald Trump has a 62% chance of winning the election, but that Kamala Harris has a 76% chance of winning the national popular vote.  They also say there is a 60% chance that whoever wins the popular vote will carry the electoral college.  That sounds like a contradiction, but isn't - the 24% chance that Trump will win the popular vote is an important component of his percentage chance of winning overall.

The national polls in this campaign have been strikingly similar to 2016 in the sense that Donald Trump has been a bit behind in most of them.  And yet in 2016 that translated into Hillary Clinton being the strong favourite on the betting exchanges, whereas this time punters are assuming that a small Harris lead in the popular vote will translate into a Trump win in the electoral college.  Could that be a case of learning the wrong lesson from history? Having been burnt before by the incorrect assumption that a Democrat popular vote win would translate into overall victory, they're assuming the same thing is bound to happen again, whereas in fact the relationship between the popular vote and electoral college numbers is much more unpredictable than that?

Well, maybe, although there has been a key change in US politics in recent years, which is that Florida has transitioned from being a toss-up state to being an almost solid Republican state.  That means more than 5% of electoral college votes, which were firmly in play for the Democrats in previous tight elections, have now been practically taken out of the game completely.  So it might not be unreasonable to assume at this stage that the electoral college is genuinely and reliably skewed in Trump's favour.  I had guessed the change in Florida was probably due to anti-communist or anti-Maduro immigrants from Cuba and Venezuela, but apparently it's more to do with anti-lockdown Republicans moving to Florida over the course of the pandemic.

Punters also have state polls with which to judge the interplay between the national popular vote and electoral college outcome, although that brings us to the issue of poll 'herding', ie. poll companies deliberately tweaking their methodology to ensure their results are similar to their competitors, to reduce risk and ensure they all stand or fall together.  That leaves open the possibility that if the polls are wrong, they could all be wrong by quite some distance.  One theory is that the overturning of Roe v Wade is motivating liberal women to vote in record numbers, and that the polls aren't picking that up.  But so far that's no more than an untested theory.

During election night in 2016 (and something very similar happened in our own EU referendum six months earlier) there was an extraordinary window of opportunity of an hour or two to make a killing on the betting exchanges, because it was obvious from the actual results that Trump was the likely winner, but there was a lag on the exchanges with Clinton remaining favourite, probably due to an ingrained belief that a Trump win was unthinkable.  Could the same thing happen tonight if Harris wins?  I doubt it, actually.  Harris may be the underdog, but nobody thinks a win for her is unthinkable, so if the early results for her are favourable, I would expect the markets to adjust very quickly.  But you never know - it's always worth just checking.