With pollsters still reeling from a presidential election in which most of them overestimated Joe Biden’s winning margins, one group of prognosticators is taking the opportunity to gloat about how much better they fared in comparison.
Betting markets had a big election season, with international bookmakers seeing record amounts of wagering on the 2020 race and an unprecedented interest in betting odds as a means of forecasting the election’s result. And now that the contest is all over bar the shouting, some betting market players are noting that they were able to call the race more accurately than the polls did.
According to online betting website US-Bookies.com, the odds-on betting favorite won 49 of the 50 states up for grabs in the election, with Georgia being the only exception. Biden surprised many observers—and most polls—by beating Donald Trump in the Peach State by roughly 14,000 votes, becoming the first Democrat since 1992 to carry Georgia.
Those same betting odds forecasted that Biden would win the election with 310 electoral college votes, according to US-Bookies, despite some polls predicting a momentous landslide that would have given Biden as many as 350 electoral votes. With the winner now called in all 50 states, Biden will end up winning the White House with 306 electoral votes, just shy of betting markets’ 310-vote projection.
Betting odds appeared to provide a more realistic reflection of the dynamics in states like Florida, which Trump won handily despite many polls giving Biden an edge going into the election. While polls had Biden ahead by roughly 1% on average in Florida, according to RealClearPolitics, oddsmakers gave Trump 8-to-13 odds—or a 62% implied probability—of carrying the state, per US-Bookies.
Likewise, in Texas, many polls indicated that Biden would be competitive in a state that Republicans have dominated for the last 40 years. But the odds gave Trump a 1-to-3 chance, or 75% probability, of winning Texas, which he eventually did by nearly 6 percentage points.
All in all, betting markets forecast a much closer race than the 7.2% lead offered to Biden by RealClearPolitics’ National Average. On November 2, the day before Election Day, Biden’s odds of winning the election were 8-to-15, giving him a relatively modest 65% chance of victory, according to US-Bookies.
As many betting and prediction market advocates claim, the value of using betting odds to predict political outcomes stems from the fact that “odds are based on people placing real money on an actual prediction,” US-Bookies notes. The idea, according to proponents, is to follow the money, as it provides a more accurate reflection of public sentiment than polling models that have appeared increasingly tenuous in recent elections.
Of course, betting odds can also prove fickle—as seen on Election Night, when Biden and Trump’s chances boomeranged back and forth as the results came in, even giving Trump an 80% chance of reelection at one point. And critics would note that the very nature of prediction markets means that they are able to hedge significantly, since they only indicate the probability of outcomes occurring—meaning that no outcome, barring one with a 0% probability, is technically ruled out.
As Lee Miringoff, director of the Marist College Institute for Public Opinion, told Fortune recently, betting odds and prediction markets “are always right by definition, even if you have a 99% chance [of an outcome].”
There are also doubts whether the very methodology of relying on betting markets to forecast outcomes can serve as an accurate predictor. “Betting markets can claim to offer a prediction, but in another sense it’s an argument that ‘XYZ’ is more likely to happen than ‘ABC,’” said Jonathan Haber, an author and researcher specializing in the field of critical thinking. “And what are the premises of that argument? That lots of people are placing bets.”
But should political betting and prediction markets continue to grow in popularity, that methodology could prove an ever more accurate reflection of public sentiment. According to US-Bookies, the record amounts wagered on the 2020 election were a key reason why betting odds for this year’s race more accurately predicted eventual outcomes than those four years ago.
“With the amount wagered on this year’s election breaking records by surpassing $1 billion, using odds to predict all sorts of political outcomes could become even more accurate if the interest in betting on them continues to increase,” a spokesperson for the website notes.
