Prediction markets may show Donald Trump with an easy lead over Kamala Harris, but the size of his lead must be taken with a grain of salt.
Presidential polls have tightened throughout the final two months of the presidential race, and prediction market prices have reflected this shift. On Polymarket’s presidential election market, Trump’s price of victory rose from just under 47 cents on Sept. 18 to just under 51 cents on Oct. 6.
On Oct. 7, one Polymarket user skewed prices in multiple markets, including the presidential victory market and the market on the winner of Pennsylvania’s electoral college votes. However, Kalshi’s presidential markets have also set Trump’s price of victory in the upper 50s. The financial movement toward Trump is legitimate.
Polls can show a tie while prediction markets simultaneously lean toward a victor. As Trump gained in voter polls, contracts on Trump’s electoral victory became more likely to pay out. It’s natural to expect those contracts’ prices to rise as demand for them increases. For example, Kalshi shows Trump ahead by 14 points, but Kalshi doesn’t have Polymarket’s price-skewing whale activity.
For astute traders, that also means that Harris’ contracts are likely underpriced and have room to rise as new money enters the market on and around Election Day. This could be a good time to buy Harris before prediction market prices regress to the mean and reflect a tighter race.
But readers and analysts who mistrust polls don’t have to take pollsters’ words for the state of the race, either. Another prediction market platform shows a functionally tied race, too.
Read more: How to Bet on Politics
Metaculus and solving the sample problem
Group predictions have four requirements before they can be considered “wise.” Prediction markets excel at three of them: independence, decentralization, and aggregation. Any trader can use their individual knowledge to inform their trades. Structured properly, no single trader can force prices to a certain value. Prediction markets are also great tools for distilling many opinions into a clear snapshot.
However, diversity of opinion is also crucial, not only for representing different viewpoints within groups, but also because one person’s error can cancel out another’s. In an election market, an extreme Trump supporter’s trade could balance out a devoted Harris backer. Across more moderate positions and error cancellations, a prediction market’s price should reflect the group’s true consensus.
Prediction markets like Kalshi and Polymarket don’t publish information about their traders that shows how representative they are of U.S. voters.
That’s where Metaculus comes in.
Weight by reputation
Metaculus is a platform with no money or fictional currency. Users register, then forecast how likely they think an event’s outcome will be. Metaculus’ twist is how it arrives at the group’s aggregated forecast. Users who have been right in the past have more weight assigned to their predictions.
Metaculus shows a dead heat between the candidates. Trump’s gains in Pennsylvania and Harris’ in North Carolina make for an uncertain race, and Metaculus reflects that uncertainty more than the real-money prediction market platforms.
Metaculus’ reputation-based system addresses the challenge that real-money prediction markets have in representing American voters. Prediction market traders are a self-selecting group willing to learn how prediction markets work, devour research extolling their virtues, and, in some cases, seek sources of information outside of mainstream sources.
That group leaves many voters out of the sample. Harris voters who still have faith in institutions may be underrepresented, but so are the forgotten men and women of Pikeville, Kentucky who are among Trump’s most loyal voters.
Weighing forecasts based on reputation is the next best thing to polls of likely voters. People who have predicted elections most accurately in the past have the most say in which direction the price moves – and which value it moves to.
If Metaculus is right, there’s plenty of value to be had betting on Harris across other prediction market platforms.
