The most nail-biting US presidential election in living memory is upon us. Many people are looking to the polls for clues on whether Trump or Biden will be victorious – but few know about prediction markets, which allow people to bet on the occurrence of events in the future: from the results of a football match to the outcome of an election.
Looking at the disparity between current polling data and what the prediction markets are saying, any politico worth their salt should take a look, or risk some nasty shocks on election night.
I work for a markets-based prediction betting platform, Smarkets. £11m has already been placed on our “Next President” market by thousands of people, from ordinary punters to experts and insiders. This makes it a highly undervalued source of information, which gives a more complete picture than polls alone when predicting election results.
“The voter approached by a polling firm has nothing to lose, so could change their mind. In contrast, traders have skin in the game: if they make the wrong call, they stand to lose cold, hard cash.”
We compared the predictions from Nate Silver’s well-respected polling source FiveThirtyEight, with our equivalent prediction market. FiveThirtyEight gives the Democratic challenger Joe Biden an 89% chance of winning the election, with Donald Trump at just 10%.
The predictive markets are notably more cautious about Biden’s chances, with data suggesting Biden has just a 67% shot compared to Trump’s 33%.
Looking at the likely winner in individual states, in North Carolina, FiveThirtyEight has Trump at 36% to win against Biden’s 64%, whereas Smarkets has the Democrats taking the state at 59%. FiveThirtyEight thinks Biden will just edge it in Georgia – whereas Smarkets thinks Trump is about 15 percentage points more likely to win.
It is clear that, although polls and markets agree Biden is the favourite, the markets see no room for complacency, and believe Trump is still very much in the race.
Polls still have their place in the debate, but prediction markets are a much more subtle tool for those who want to understand the full picture.
Polls are a snapshot of public voting intention at a particular moment in time. Most people the pollsters speak to have no special insights, other than knowing their own mind. In contrast, prediction markets are composed of a group of highly interested people trying to correctly forecast a result in real time, striving to keep up with the wealth of new information becoming available every day.
The voter approached by a polling firm has nothing to lose, so could change their mind, or give the pollster an answer they aren’t totally sure of. In contrast, traders have skin in the game: if they make the wrong call, they stand to lose cold, hard cash.
“The current discrepancy between the polls and the markets suggests something may be off.”
While it is true that polls can sample a wider range of people than the self-selecting traders on political markets, and then weight the results to produce a balanced representation, incorrect weighting produces skewed results – as in 2016 when pollsters underestimated Trump’s support from blue collar voters.
We will soon discover whether they have been successful this time round, but the discrepancy between the polls and the markets suggests something may still be off.
That’s not to say polls are irrelevant, but they are just one source among many used by traders on betting exchanges, who evaluate and interrogate polls, rather than using them unquestioningly as many political pundits seem to.
Another weakness of polling data is that it only looks at voting intention, whereas market participants are also thinking about issues such as voter suppression, irregularity in counting, turnout, disqualification of postal votes and a candidate’s performance in debates. All these factors could affect the outcome of the election: many of those polled may not have their vote counted after all.
But it’s a mistake to conflate market traders with traditional experts, much-maligned as an out of touch political class who consistently underestimate the underdog, from Corbyn to Farage. Prediction markets depend on the wisdom of crowds. The more individuals that participate, the more data there is, and the more effective the predictions.
Studies show that when a diverse group of people come up with an answer – about anything, from the weight of an ox, to the performance of stock markets – it is likely to be better than that of the smartest person in that group, or even a group of experts. Our market has many thousands of participants putting their money where their mouths are, from campaign insiders to everyday punters.
Perhaps the most valuable “lifeline” on Who Wants to Be a Millionaire? is to ask the audience. For anyone anxious about the result, it is well worth looking to the markets for an unorthodox view that just might prove accurate.
Sarbjit Bakhshi is Head of Political Markets at Smarkets