In the weeks leading up to the 2024 presidential election, many thought the contest would be one of the closest in recent memory, with most polls showing a toss-up race between Donald Trump and Kamala Harris. The Economist notably listed a 56% chance of Harris defeating Trump in its final election projection, which relied heavily on traditional polling data. Ultimately, however, the election was not as close as many predicted, with Trump defeating Harris 312-226 by electoral vote count and 50.2% to 48.1% by popular vote count (as of the time of this writing). The discrepancies between polling data and the results of this and other recent contests beg the question: Is there a better way to predict the outcome of elections? Enter the betting markets…
While betting directly on election outcomes is new, indirect betting is not. In 2016, Sam Bankman-Fried (yes, that Sam Bankman-Fried) and Jane Street, a global proprietary trading firm, built robust models that incorporated data all the way down to the county level to help predict the outcome of that year’s election. While the firm could not bet directly on the outcome of the race, it could take positions that the team believed would be profitable based on the model’s projected result. To that point, Jane Street was able to front-run election updates from major media outlets and build a short position in the S&P 500 Index, as the team thought a Trump victory, which their models projected far before the mainstream press, would be negative for markets. While this trade initially showed signs of promise, it ultimately led to some of the biggest losses in the history of the firm as equities turned positive in the wake of Trump’s first electoral victory.
This presidential election cycle provided more opportunities for speculators to bet directly on the outcome of the race, with odds updating by the minute. Shortly after the first presidential debate on June 27, for instance, betting markets exhibited much higher odds of Harris winning the presidency than then-nominee Joe Biden. Then on July 21, Harris replaced Biden at the top of the Democratic ticket, a move forecasted weeks earlier by betting markets. While Harris surged in betting markets in the weeks following this change, Trump became the odds-on favorite to win the contest in the days leading up to election day, with a 54% chance of taking the presidency. Betting odds started moving quickly last week on election night when results began pouring in. Roughly 40 minutes after the first polls closed, betting markets began shifting heavily toward Trump despite the electoral count being just 23-3 in his favor (with 512 electoral votes outstanding). Harris was a longshot bet in a matter of hours despite still having several paths to victory, as betting markets indicated a 95% chance of a Trump victory before midnight. The Associated Press finally called the election for Trump at 5:34am.
Going forward, odds markets may be better predictors of election results than more traditional polling data. This is due to the wisdom of crowds, the incorporation of extensive data in odds calculations, and the fact that people tend to be more honest when betting than they are with pollsters.