Could betting markets really predict the presidential election results better than polling?

Could betting markets really predict the presidential election results better than polling?
Source: Mike Blake, Reuters
While the oddsmakers provide yet another frame for predicting the outcome of presidential elections, there is little evidence they are more accurate than polling and other predictive methods.

With the United States’ presidential election less than a month away, there is an increased focus on the polling, much of which favors former Vice President Joe Biden over the incumbent, President Donald Trump. Yet, following Trump’s improbable win in 2016, literally overcoming the odds, many election watchers are hesitant to trust the polls this year.

But election speculation remains a popular hobby, one that has been turned into a lucrative activity. Betting markets, which generally focus on sporting events, now offer odds on the election results, turning the pastime of political forecasting into a moneymaking endeavor for gamblers.

At times throughout 2020, the betting sites have been at odds with pollsters. In recent months, though, oddsmakers, mimicking the polls, have largely coalesced around the assumption that Biden is favored to win. It’s possible the betting markets are a valuable gauge of politics at any given moment, but skeptics feel gambling odds are far too volatile to accurately predict election results.

The betting markets

There are countless gambling sites that focus on sports betting, among them the popular PredictIt and SportsBettingDime. These betting markets – or prediction markets – have proliferated in recent years, especially following a Supreme Court decision in 2018 that overturned the Professional and Amateur Sports Protection Act of 1992.

For both casual and regular gamblers, betting on sports is just one of many gambling options. Betting on political outcomes has been popular for decades, though it wasn’t until 2008 that political betting markets entered into the mainstream. That was when Intrade helped popularize the concept.

Though the site pulled down its trading platform in 2013, in both 2008 and 2012, it was among the most talked-about markets for betting on the presidential election. Intrade didn’t just present itself as an alternative to polling, it argued that its methods were in actuality more accurate.

“Because trading is incentivized by real money,” the website currently reads, “Prediction Market forecasting is almost always more accurate than traditional forecasting methodologies. In the simplest terms, Prediction Markets collect and aggregate the ‘Wisdom of the Crowd’.”

In 2008, the final election prediction on Intrade was within one electoral vote of the final results. Intrade predicted then-Senator Barack Obama would defeat Senator John McCain with a 364-174 electoral college lead. In fact, the final total was 365-173 for Obama.

In 2012, Intrade correctly predicted Obama’s victory once again, but undershot his electoral college total by 29, predicting 303 votes to his actual 332.

While those are strong results for Intrade, it is worth noting that pollsters were also predicting a victory for Obama in 2008. Furthermore, in a Gallup poll of who Americans thought would win the election in 2012 (as opposed to who they wanted to win), respondents favored Obama over his opponent, Senator Mitt Romney.

In 2016, when the polling overwhelmingly indicated former Secretary of State Hillary Clinton would easily win the election, PredictIt’s betting market largely aligned with the polling. On November 7, the day before the election, PredictIt favored Clinton to Trump 4:1. In the end, despite the long odds, Trump pulled off the victory.

All of this suggests there is no clear edge in methods for predicting the results.

What are the odds?

This year, both PredictIt and SportsBettingDime are, yet again, predicting a Trump loss. The two sites’ betting platforms are slightly different, though.

For PredictIt, their system is set up to resemble a stock market. A bettor can buy “stocks” for who they believe will win the election. The greater the odds that one candidate will win, the more expensive their stock is. So, for instance, on October 8, Biden’s stock hovered between 67 and 68 cents, whereas Trump’s was 36 to 37 cents.

Other long shot bets for the presidency include both vice-presidential candidates, Senator Kamala Harris and current-Vice President Mike Pence. Further down the list are Hillary Clinton and a range of other politicians and public figures, including Facebook creator Mark Zuckerberg and rap and fashion mogul Kanye West.

Similarly, SportsBettingDime reported on October 6 that betting markets only gave Trump a 38.3% chance of winning, versus Biden’s 61.7% chance. SBD’s betting system is a bit more complex than PredictIt, relying on a +/- betting odds system that is based around US$100 bets. All odds are given in three-digit numbers.

So, for instance, on October 6, Biden was rated at -179 and Trump was rated at +151. What that means is a person betting on Biden would need to place US$179 to win US$100. Conversely, a person betting on Trump would win US$151 if they bet US$100.

These betting oddsmakers do not limit themselves to just the presidential race. Prior to the vice presidential debate on October 7, SportsBettingDime featured a range of odds on potential events related to the debate.

Bets included “Which candidate will lie first (per NYT fact-checker)?” (lower odds were put on Pence, meaning he was expected to lie first) and “Will either candidate test positive for COVID-19 by Oct. 13th?” (low odds were on “No”).

Are the betting markets good predictors?

While the oddsmakers provide yet another frame for predicting the outcome of presidential elections, there is little evidence they are more accurate than polling and other predictive methods.

Even the betting sites themselves seem skeptical of how accurate the markets actually are.

“It is baffling to me how drastically the odds to win the 2020 election shift from month to month,” Sascha Paruk, the editor of SportsBettingDime, told TMS. “Since the start of 2020, alone, Donald Trump’s chances to win re-election have swung from -214 (68.2% probability) to +151 (39.8% probability). It is remarkable how short-sighted both bettors and oddsmakers seem to be, thinking that what happens in February will still be a prime factor in November.”

Paul, a regular user of betting sites, also told TMS that he is skeptical betting markets do more than provide a snapshot of what people think will happen:

“It can’t be denied that both the betting markets and the pollsters got [the 2016 election] wrong. I wouldn’t say I trust betting markets more than polls, but where the polls ask a sample and then extrapolate an outcome, I think the betting markets are interesting to follow because they reflect public opinion on what a likely outcome will be.”

November 3 will provide another opportunity for the betting markets to prove their worth. However, no matter which way the votes go, if the markets continue to align with the polls – which have improved for Biden since Trump was diagnosed with COVID-19 – they aren’t likely to convince naysayers that they are a better prediction tool than traditional polling.

Have a tip or story? Get in touch with our reporters at