The US election is a chance for Australians to engage in their favourite pastime — gambling. Betting odds are often used as a basic indicator of predicted success in sporting contexts, but in politics, that’s not always the case.
In politics, the convention is an altogether more scientific method of polling and mathematical modelling. One would think this methodology would yield more accurate results, but on election night in America, it was the punters who were right and the pollsters who were wrong.
Host of the popular The Rest is Politics podcast Rory Stewart, a former Tory MP, predicted that Kamala Harris would win solidly. Allan Lichtman, the historian known as “Nostradamus” and who has predicted nine of the 10 past elections, was also wrong in backing the vice president. Prominent pollster Ann Selzer sent waves through the American political establishment after predicting Harris would win Iowa — a traditional battleground (but recent safe Republican state) — in what would have been an 11-point swing towards the Democrats from the past presidential election. Iowa ended up solidly won by Donald Trump with 56% of the vote.
Conversely, on election eve, Australian bookmaker Sportsbet had Donald Trump listed as the favourite to win the election. So how did it get right what the pollsters got wrong?
Crikey tracked the betting odds for a Kamala Harris victory as election results rolled in across America, with interesting results. While Sportsbet did not respond to our request for odds data in time for publication, we found enough readers willing to share their bet slips.
Betting odds in Australia are commonly expressed as a dollar figure — i.e. if an outcome is paying $2, one would receive $2 for every dollar staked.
On October 7, just under a month out from election day, odds.com.au reported Sportsbet was paying $1.91 on Harris to win the presidency, level with eventual victor Donald Trump. By election eve, that was up to $2.32 per News Corp reports, with Trump a $1.62 favourite.
At 10.50am AEDT on election day, November 6 — just under an hour after the first polls had closed — Sportsbet was paying $2.30 on a Harris win. The first state to be called for Trump, per Associated Press, was the safe Republican state Indiana at 11.05am. At approximately 12pm, Kamala’s price was up to $3.10, which slowly crept up over the course of the hour.
By 1.28pm, 20 minutes after AP had called traditional bellwether state Ohio for Trump, Sportsbet was paying $3.55 on a Harris win. By 2pm, following a flurry of safe seats falling to Trump, Harris had blown out to around $8, with one punter reporting to Crikey they had placed a bet of $10 at approximately 2pm. Former swing state Iowa fell to Trump at 2.40pm, and by 2.56pm, Harris sat at $12.
Notably, battleground state Virginia being called for Harris at 3.42pm appeared to cause a momentary drop in odds for the Democratic candidate. One punter reported Harris paying $31 at approximately 4pm, but another provided a betslip to Crikey that was paying $17 at 4.49pm. By 5pm, swing state Georgia had been called for Trump and Harris was back to paying $34. At 6.33pm, Trump took the stage in West Palm Beach having been all but confirmed as the victor of the election.
Betting odds can be a useful predictive tool that can add nuance to traditional polling methodologies, according to University of Sydney academic Dr Francesco Bailo, a lecturer in data analytics. Bailo said the betting markets can only reflect the information consumed by the public and are also tarnished with additional factors that can colour results.
“The people in these markets consume the same information everybody has consumed, so basically they will reflect more or less the polls and current understanding about the election,” Bailo told Crikey.
Despite these limitations, Bailo said that while it was “really difficult to exactly track and aggregate every market”, the major markets were “pretty in line” with polls, which predicted Trump had a greater chance of winning the election.
Asked about the phenomenon of the “silent majority” and polls underestimating Trump, Bailo told Crikey that there was “a clear underestimation of Trump … this is one of the most interesting (things) … something we need to really reflect on”.
“In a broader sense, there is something we’re not capturing when we try to describe Trump,” he said.
“Maybe the categories we’re using — male, female, educated, non-educated, white, non-white — are not capturing the entire thing.”
Asked whether there was anything to change in terms of accounting for that underestimation, Bailo said, “It was good to have an idea of how you want to integrate other sources (of data). I think the betting market might be one. If we have a good understanding of (bettors), we can probably be able to aggregate the betting into our models to better improve the accuracy of our polling.”
Anyone affected by problem gambling can get immediate assistance by calling the National Gambling Helpline on 1800 858 858 for free, professional and confidential support 24 hours a day, seven days a week.
