Even during the last five years of dramatic political betting, I cannot recall anything quite like the last week. An avalanche of money poured on Donald Trump to be re-elected - on an unprecedented scale for this stage of the contest - dramatically altering the odds.
Scale of Trump gamble is unprecedented
Heading into Labor Day weekend - the traditional starting gun for the final, intense stage of the election - Betfair’s live exchange market implies a race that is too close to call. Trump’s chance, based on their odds, rose from 35% to 46% during August. The firm reported an average above £1M per day traded on their market for Next President - the majority for Trump.
I think the market move was driven by a combination of the factors discussed previously on these pages. First, Trump supporters feeling pumped after what they regarded a successful convention. That exacerbated other market characteristics, such as - gender bias among gamblers and Trump supporters. A global trend of ‘Alt-Right’ gambles. Trump’s particular fame factor. Fears the election will be rigged.
Second, a narrative built on an untested assumption that the violence in Kenosha and Portland would work to Trump’s advantage. Third, relentless propaganda regarding the violence, Biden and fake polls.
Gap between odds and polls has never been wider
The effect of these last two, plus lack of trust in the process, are legitimate warnings against complacency for Biden backers. They do not, given everything else we know, amount to anything like a 46% chance. There is no reason to distrust the polling models measuring his chance around 30%, at best.
Let me restate important facts with regards 2016. Trump lost the popular vote, with 63M in total and 46.1% share. That wouldn’t usually be a big enough share but that unique election between two unpopular candidates saw third parties take out 5% of the vote, lowering the bar. He won the electoral college by edging three key states by a combined margin of just 77K votes.
On the same day, Republicans won the nationwide House of Representatives by a 1% margin, scoring 3% higher than Trump. Turnout among key Democrat-leaning segments - black and young voters - was depressed. In other words, Republicans were the more popular party, better motivated and therefore benefited from differential turnout. Yet Trump still only won by the narrowest of margins.
Trump needs to improve on 2016 amid worse conditions
To win again, there is no margin for error. Actually he needs to improve on that performance, amid less favourable conditions. Because there is no strong third party challenge this year, I estimate he needs at least around 47% nationally. Given that turnout soared mid-term - to 50% at the 2018 mid-terms compared to 36% in 2014 - that target will be higher than 63M votes.
For example if turnout were to rise by a tenth to around 61% overall, Trump would need to gain around 7M extra voters. Highly unlikely, given that the increased turnout seen since 2016 has mostly benefited Democrats and come from voter segments likelier to lean blue - women, minorities, young.
Liberal enthusiasm has been evident across the board - whether special elections, primaries or even for the Supreme Court in marginal Wisconsin. The latest registration numbers for Pennsylvania show the Democrats enjoying an edge over Republicans of 132,000 in newly registered voters - three times the size of Trump’s margin of victory in that pivotal state.
That advantage can be seen in state after state. It isn’t surprising, given Trump’s unpopularity with young voters. Plus the numbers are probably a lot worse than the simple Democrat/Republican split. The largest number are Independents - a group that splits against Trump.
Trump and his party have regressed across the board since 2016
Indeed, there is no evidence whatsoever that Trump or the Republicans have progressed among voters since what seems like a peak moment in 2016. Trump’s approval rating fell immediately after inauguration from an already historically low mark and has been entrenched below 45% ever since.
From winning the House by 1.1% in 2016, Republicans lost by 8.4% in 2018. Generic Congressional polls have consistently confirmed that picture before and after those mid-terms elections.
Also, of course, there are head-to-head polls. Biden’s clear lead isn’t new. The pair have been polled frequently ever since Trump entered politics, because the former VP was regarded a potential rival. Then as now, Biden fared significantly better than Hillary Clinton, often leading by double-digits.
To back Trump, we would be betting that all these polls are fundamentally wrong and have been, consistently, for years. That he has an army of supporters who didn’t vote for him amid the optimism of 2016 but will do now. Or that there are sufficient Biden voters who can be won over, despite only around 7% of such ‘persuadables’ being recorded by polls. The US electorate is extraordinarily polarised and therefore predictable.
Sure, polls can be wrong and contain a margin of error. Sampling can be slightly wrong. It is very hard for them to identify differential motivation and turnout. Both are explanations for their slight underestimation of Trump - by 1.1% nationally but slightly higher in swing states.
The sampling error has been corrected in accordance by the top pollsters. As for mismeasuring differential turnout, I’d regard that much likelier to favour the Democrats this time. I suspect pollsters may have trouble picking up newly engaged minority and young voters.
Entering this final stretch, I was perfectly happy with the position built up since 2016 - an average lay (oppose) of Trump at odds of 2.22. No plans to cash out. However once he shortened to just 2.06 on Betfair, I had to lay some more. These odds are completely wrong and the election looks ever more like a historic opportunity to find value.