Not so long ago - four years to be precise - there was a growing belief that betting markets offered the best guide for predicting elections. I conducted literally hundreds of interviews on the subject.
Within the space of a few mad months in 2016, the theory imploded, following historic upsets in the two biggest political betting markets of all time. How does it stand up after gathering four more years of evidence? Should pundits be following the betting with a laser-like focus or ignoring it? Should gamblers be backing favourites or looking for outsider punts?
Pre-2016, Political Favourites Were Ultra-Reliable
First we must remember why the theory developed. Political betting markets only grew to large global scale in the 21st century. Betfair - the world's first online betting exchange - swiftly became the premier platform after being founded in 2001. This is where the biggest amounts were traded and - driven by peer-to-peer betting as opposed to the opinions and commercial imperative of an individual bookie - offered the clearest guide to crowd opinion.
Elections in the UK and USA produced by far the biggest markets. In every general election in either country between 2001 and 2016, the favourite at one hundred days out went on to win. By favourite, I mean to win the most seats in the UK parliament or to win the presidency.
There were caveats. This amounts to a small sample - six elections in total. The secondary market in both the 2010 and 2015 UK elections - regarding whether one party would win an overall majority - failed to meet the 100 day criteria. The 2004 US election was indeed won by George W Bush, but betting signals earlier on election night swung towards John Kerry, implying up to an 80% chance.
More generally though, the betting did provide a great signal during that period. Smaller markets - London Mayor, German Chancellor, French President, Scottish Independence Referendum, US mid-terms - met the criteria.
Brexit and Trump Broke the Mould
That ‘wisdom of crowds’ hypothesis collapsed in 2016. In the UK's referendum on EU membership, Leave only became favourite once the result was almost mathematically certain. It was a 10-1 chance as the first results emerged. So too was Donald Trump, early on election night.
Since those seminal moments, the picture has been mixed. There have been two further UK general elections. The perfect run of favourites in the ‘most seats’ market was maintained by the Conservatives winning in 2017 and 2019, but that secondary ‘overall majority’ market flipped late again. The 2017 election turned out to be particularly unpredictable. In the USA, the Democrats did justify long-term favouritism at the 2018 mid-terms.
Results elsewhere have also proved inconclusive. For example, the Social Democrats led all the way in Sweden’s 2018 general election but Scott Morrison’s Liberal Party pulled off another shock with a late swing in the 2019 Australian Federal Election. When Emmanuel Macron became French President in 2017, his new En Marche party had started out at 66/1 and were only rated third in the betting at the hundred day stage.
Early Betting, as Opposed to Polls, Points Towards Trump
What therefore, are we to make of this theory heading into this year’s US Election? We are still around two hundred days out so plenty can change. However right now, there is a vast disparity between what the polls imply and the weight of money in betting markets.
The betting remains relatively close between Donald Trump and Joe Biden - 2.0 versus 2.36 on Betfair. The party split is virtually even at 1.97 for Republicans versus 2.04 for the Democrats.
Polls, however, clearly point towards the challenger and have done consistently, for several years. When he was Obama’s VP and a mooted contender for 2016, pollsters began comparing Biden against Trump. He consistently led by double-digit margins and the trend continued after Trump took office.
For example, the last ten polls collated by RealClearPolitics during 2018 showed Biden leading between 9% and 18%. Among 90 head-to-head surveys taken towards that average since the beginning of 2019, Trump has led in only four. The more recent average is at least moving his way at 5.9%.
Other polling indicators are grim for the incumbent. Trump’s approval ratings have been entrenched underwater in the mid-forties since taking office. The generic congressional ballot shows the Democrats consistently ahead by an average above 8% - in line with the ‘Blue Wave’ that powered their best mid-term result since the 1970s.
State polls are of particular importance, because Trump demonstrated in 2016 that his support is more efficiently spread around the key states. However they don’t read much better. The latest Arizona numbers, for example, show Biden 9% ahead. This is a state where Trump beat Clinton by 4%.
I’d say two factors explain the differential. First, whereas polls are only ever a snapshot of public opinion, gamblers are trying to predict how the future will pan out. Events, scandals can transform the dynamics. Many doubt Biden will be effective on the campaign trail. We are all guessing regarding the extent and impact of coronavirus.
Does Betting Reflect Divisiveness of Unique Trump Era?
More widely, betting during the Trump era has been less rational and more partisan than any in living memory. Trump is a uniquely polarising figure. It has become near impossible to hold a neutral stance about him and that applies to gamblers or the media from which we all derive our information.
The betting on those mid-term elections never seemed particularly rational, nor whether he’d be impeached by the House of Representatives or on his exit date.
For example around 3.5 was available just a few days before impeachment became almost inevitable. Alternatively anti-Trump punters lost plenty backing him to leave office in 2017 and 2018 at odds below 4.0.
Whilst the ‘wisdom of crowds’ theory may turn out to be of some value over the long-term, across numerous countries, I’m sceptical it applies to the USA during this particular era.
My view is that we can only place any real store in the most respected polling companies - Fivethirtyeight’s rankings offer an excellent guide - and how they see the swing states that will determine the electoral college. I am sceptical that the market will respond accordingly though. There will be no shortage of ‘alternative facts’ and polls available on the internet.