There has never been a mismatch like it in the history of political betting. According to Betfair - based on peer-to-peer trading and therefore the ultimate guide to market sentiment - Joe Biden is 57% likely to win the election, compared to 41% for Donald Trump. Polling models, however, paint a very different picture.
 
The Economist/Yougov rate Biden 88% likely. Fivethirtyeight are a lot more circumspect, factoring in a greater likelihood of change during the campaign, yet still project significantly higher than the betting at 72%. I can think of least eight possible explanations.

1: Distrust of the polls

This narrative developed after 2016 but it doesn’t really stack up. The final result was only 1.3% more favourable to Trump than the RCP average - easily explained by late defectors from third parties and differential turnout. Polls in key states were further out but not on the scale required were the election today. At the biggest election since - the 2018 mid-terms - the polls were spot on.

2: Fear of a 2016 repeat

Polling error or not, it is nonetheless true that Trump’s win was massively against the odds and expert consensus. Gamblers are understandably wary of a repeat but, as explained recently, the conditions are very different this time. Biden’s poll lead is higher, more consistent and he is nowhere as toxic as Clinton. There is no strong third party splitting the anti-Trump vote.

3: Lack of confidence in Biden

Perceptions can of course change during the campaign, especially about the much less defined Democrat. Keeping a low profile from his basement has worked a treat for Biden, letting Trump damage himself over Covid, but that will change in September. He will be exposed to scrutiny like never before and many doubt he’ll pass the test.

The campaign will indeed be challenging for a 77 year-old, whom Team Trump say has dementia. Much earlier presidential bids were ruined by gaffes. However he may defy low expectations at set-piece moments. In the last two Democrat primary debates, Biden was widely deemed the clear winner. Trump also lost all three debates in 2016 by big margins.

4: Scandal could be a gamechanger

US elections are incredibly dramatic affairs. Scandal and smear campaigns are guaranteed. 2016 saw two ‘October Surprises’ within minutes of each other - the Pussygate tapes and Wikileaks’ release of the Podesta e-mails. Trump has already been impeached for bullying the Ukranian government over military aid in exchange for manufacturing dirt on Biden. 

His team are openly pursuing the same tactics and flagging it up on their media. We will see what emerges and any effect. However scandal could just as easily hurt Trump, especially regarding his long-running legal problems and ongoing revelations about his ties to Russia. 
In either case, the impact needs to be big as polls show very few persuadable voters.

5: Fear of voter suppression or vote-rigging

This is the likeliest explanation for Trump’s recent betting momentum. His placeman’s sabotaging of the postal service has made big news and the tactics are barely denied. Trump is determined to restrict mail-in voting. There will be further voter suppression tactics. 

Mail-in votes will take a long time to settle, setting the stage for chaos. It doesn’t take a wild imagination to envisage Trump refusing to accept the result and a dangerous stand-off. I do fear the worst on that front but the betting will be settled on votes, regardless of whatever chaos ensues. 

I think the postal vote suppression will backfire. It hasn’t and can’t be disguised, creating resistance from postal workers and terrible publicity. It hands the Democrats a powerful narrative and will make people more determined than ever to vote, irrespective of Covid risk and long lines. 

As for vote-rigging, there is no evidence that Russia have the capability to do so. Their successful interference in 2016 was via hacking and propaganda. They were highly active at the 2018 mid-terms too, but it didn’t prevent Democrats winning, bang in line with the polls.

6: Gender Bias

Could this be a factor? Trump’s vote overwhelmingly skews towards men. So are gamblers. Partisanship is at unprecedented levels, amplified by social media. Are people simply backing their team?

7: Alt-Right candidates usually inspire gambles

If so, it is worth noting that gender bias also applies among supporters of ‘Alt-Right’ candidates in other countries, who have repeatedly been the subject of vast, irrational, failed gambles since 2016. Geert Wilders in Holland. Marine Le Pen in France. Sweden Democrats and their smaller rival the AfS. Trump is more famous and this market is bigger than all of them combined. This is a global gamble and the scale of it may well hold the odds up. 

8: The celebrity factor

Fame is at the heart of it, in my view. The moment Trump entered the Republican primary in 2015, public engagement soared like never before. He has a celebrity brand that dwarfs any politician and a media presence that sucks all the oxygen out of the room. Love or hate him, everybody has an opinion.

It means the election will be framed as a referendum on him. That is a negative based on entrenched long-term polling trends but, as a narrative, it cuts through to the casual observer. Backing Trump at odds-against, to repeat the upset that he pulled off in 2016, has romantic appeal to a gambler. Backing his relatively low profile opponent at odds-on doesn’t.