Online betting has been around for years. Platforms like Polymarket and PredictIt operate similarly to sports betting apps like DraftKings, but they allow users to bet on nearly anything from earthquakes, the creator of Bitcoin, or how many times Donald Trump might say “God” during a speech. The underlying mechanism is simple: participants buy and sell shares based on which outcome they believe is more likely to occur. As more participants buy shares, the price, or odds, of that outcome increases, reflecting a higher probability of occurrence. If a trader’s prediction proves correct, the value of his shares increases to $1, resulting in a payout. If wrong, the value drops to zero.

Until recently, political betting markets were largely illegal in the United States. The ban dates back nearly a century, primarily due to regulations designed to prevent financial speculation and gambling on public events. Platforms like PredictIt and Polymarket managed to operate by being based offshore or through legal loopholes, such as academic affiliations or cryptocurrency-based frameworks. However, just this last month, a Washington, D.C. court gave the green light to U.S.-based startup, Kalshi, a betting platform that allows users to wager on various outcomes, including election results. However, it is restricted to U.S. participants.

This year, one of the most active markets is predicting the outcome of the 2024 presidential election. On Polymarket, a decentralised, cryptocurrency-based platform and one of the world’s largest prediction market, betting volumes surged past $2 billion. This past week, Donald Trump’s lead over Kamala Harris reached a new high. Although these markets are known for their constant fluctuations, Polymarket currently shows the former president with a 61.4% chance of victory, while Harris trails at 38.6%, at the time of writing. The prediction platform also provides insights into swing states that could play a key role in the election outcome. Trump is currently leading in several of these states: Arizona (68%), Georgia (65%), Pennsylvania (59%), Michigan (57%), Wisconsin (56%), and Nevada (50%).

Traditional electoral polling provides a structured method for understanding voter sentiment. These polls typically involve surveying a representative group of individuals, asking them who they intend to vote for or how they view key issues. Pollsters then adjust the raw data to account for factors like demographic groups, regional distribution, and historical voter turnout patterns to produce a forecast of the overall electorate's leanings.

Models from Reuters, CBS News, FiveThirtyEight, Harvard-Harris, and The Economist all show a slight edge for Harris. One of the most popular poll-aggregating sites is RealClearPolitics, which compiles state polls into a simple moving average. Their latest national poll tracker indicates 49.2% of voters support Harris, compared to 47.5% for Trump. In key swing states, RealClearPolitics' averages show Trump with a slight lead in six out of seven states, while 538’s averages, led by Nate Silver, place Harris ahead in four of those seven.

While national polls give an indication of popular support, they do not account for the U.S. electoral college system, which ultimately decides the presidential election. Each state’s electoral votes are determined by the number of its senators and representatives in Congress, with Washington, D.C., receiving an additional three electoral votes. Except for two states, a winner-takes-all system applies, meaning the candidate with the most votes in a state wins all of its electoral votes. To secure the presidency, a candidate must reach 270 electoral votes. In the event of a 269-269 tie, the House of Representatives steps in to decide the winner through state delegations. 

The Nate silver Bulletin model is considered often more accurate than traditional poll data alone because it goes further by weighing pollsters based on economic indicators, their track record, and campaign fundraising data.

Source: Silver Bulletin

Betting markets and traditional polls offer two distinct perspectives when it comes to predicting election outcomes. Unlike polls, betting markets are dynamic, similar to financial markets. While polls offer periodic snapshots of voter intentions, betting markets can reflect real-time new information, often providing a quicker measure of how breaking news, such as debates or scandals, impact the race. “The markets are tending to be about a week ahead of the polls,” notes Harry Crane, a statistics professor at Rutgers University. Betting markets also integrate a broader range of information, including voter turnout projections, campaign strategies, and news coverage, aggregating it all into one probability estimate. As Crane puts it, “The poll is asking you who you’re going to vote for. What people care about in betting markets is who’s going to win.”

Another key distinction is the economic incentive driving betting participants. While bettors may hold personal political opinions, their primary motivation in betting markets is profit, as they are risking their own money. This financial stake encourages them to analyse all available information, including polls, news, and expert opinions, to make the most accurate prediction possible. Harry Crane studied the accuracy of platforms like PredictIt and explains: “Any given piece of information that you might have, such as a poll or an aggregator, or news, or a pundit’s opinion, all of that is being combined in the participant’s markets who have an incentive to get it right and to make money off of it.” This economic motivation makes bettors more likely to factor in variables that polls might overlook.

Source: Comparison between the prediction accuracy of 538 and ElectionBettingOdds from 2016 to 2020, from MaximumTruth