A Look at Polymarket, the New and Intriguing Phenomenon for Predicting Election Results Using Bets

Polymarket’s forecast on Oct. 25 for the U.S. presidential election aligned perfectly with the official results days later. The prediction of President-elect Donald Trump’s landslide victory over Vice President Kamala Harris was remarkably accurate. However, there’s much more to this prediction platform than what you see at first glance. What exactly is Polymarket, and how does it differ from traditional polls and expert forecasts?

Prediction markets. Together with Kalshi and PredictIt, Polymarket is one of the leading prediction markets. These platforms allow users from around the world to bet real money on the outcomes of various events, including elections. Users can buy “shares” priced between $0.01 and $1.

Share prices reflect users’ opinions about the likelihood of different outcomes. For example, if Trump’s shares are trading at $0.70, it suggests that there’s a consensus that he has a 70% chance of winning. Users can then place bets on Trump or Harris, which alters the share prices and overall sentiment.

The basis of prediction markets. Proponents of platforms like Polymarket argue that requiring users to purchase shares (essentially betting) and the fact they have a large number of users make the system more accurate than traditional polls, where respondents can easily lie.

The theory is straightforward: If a certain outcome appears likely, investors will work to ensure that the shares retain their value because they can profit from accurate predictions. Additionally, investors leverage information from polls and analysts to inform their bets, which theoretically enhances the validity of their predictions. Those who make correct predictions earn $1 for each share, while those who don’t lose their bets.

A crypto-based infrastructure. Polymarket, for example, operates on the Polygon network, a technology built on ethereum’s proof-of-stake blockchain. All transactions are conducted using USDC, a stablecoin pegged to the U.S. dollar. This helps avoid volatility and keeps transaction fees very low.

Polling and election forecasting. Traditional methods of predicting election results have primarily relied on two main groups:

1. Polls. The most well-known method is simple. It involves asking a group of people (known as the “sample”) about their voting intentions. The results from these surveys are then extrapolated to represent a larger population. However, polls don’t assign probabilities to the election results. They simply indicate the number of people supporting a candidate at a specific time. While polls can be useful, they can also fail dramatically.
2. Expert election forecasts. Platforms like FiveThirtyEight and The Hill and analysts like Nate Silver (who founded FiveThirtyEight) are well-known for their election forecasts. Entities and experts leverage surveys and additional data, such as economic indicators and historical trends, to predict voting intentions and assess the likelihood of one candidate defeating another. Silver gained recognition for accurately predicting Trump’s victory in the 2016 election.

Too much volatility. Polymarket’s forecast made on Oct. 25 was almost negligible. The platform claimed that its system had accurately predicted the election outcome at that time. However, the predictions were continually changing. Initially, the predictions had shifted slightly in favor of both Trump and Harris. Afterward, they also changed, although they consistently favored Trump. In fact, just one day before the election and in the days leading up to it, Polymarket’s forecast closely mirrored the final result. This volatility is a notable aspect of prediction markets, influenced by various factors that can affect predictions.

Freddi9999. This is the alias of a French investor who placed millions of dollars on the bet that Trump would win the election. This bet turned out to be very profitable for him, with Bloomberg reporting (based on data from the crypto consultancy Chainalysis) that his earnings could have reached $85 million. However, a massive bet like this raises questions about the reliability of these predictions because it highlights how “whales” in the prediction markets can amplify the views of speculators. Their influence can affect the final outcome by shaping the betting behavior of other users.

Advantages of Polymarket and Kalshi. These decentralized systems can provide valuable information and can react more quickly to various events during the election campaign. According to The New York Times, this was evident in June, following a debate where President Joe Biden appeared quite erratic.

While polls were slow to respond to calls for him to withdraw from the race for reelection, Polymarket reflected a significant shift in sentiment. Before the debate, stocks betting on his potential withdrawal were valued at $0.21. The morning after the debate, that figure jumped to $0.43, and a week later, it rose to $0.60. Eventually, Biden announced his withdrawal a month after the debate.

It’s worth noting that rapid reactions to events can contribute to the volatility of predictions.

Regional and monetary limitations. In the U.S., betting on electoral processes is illegal, which is why Polymarket was only allowed to operate outside the country. However, it’s unclear if U.S. citizens have used VPNs to participate in these markets. For its part, Kalshi was exceptionally allowed to operate during these elections, enabling citizens to place bets there.

Additionally, Polymarket doesn’t limit the amount of money users can wager, whereas PredictIt sets a maximum bet of $850.

Which system is more accurate? At least in this election campaign, Polymarket has emerged as one of the indirect winners. Its predictions, along with those from Kalshi and PredictIt, were closer to the final outcome than polls and analyst forecasts, which indicated much tighter victories for Trump. Nonetheless, whether predictive markets are superior to other forecasting methods remains uncertain.

Image | Phillip Goldsberry

Related | FTC Ban on Fake Reviews Goes Into Effect, Authorizing Agency to Fine up to $51,000 Per Violation