Paradigm February 2026 Poll on Prediction Markets
03.12.2026|Dominique LittleJustin Slaughter
Over one-third of voters already use prediction markets, whether it's to check the forecasts or trade.
That’s the top finding of our new poll on prediction markets, and yes, this was a surprise even to us.
To be clear, prediction markets have already proven themselves to be a meaningful part of how Americans understand politics, sports, and the economy.
From the way they are being used to monitor political campaigns and the economy in real time to the increasing intensity of debate over how to regulate them, prediction markets have proven themselves as one of the most significant developments in financial markets this century.
But while the political debate over prediction markets grows louder by the week, one critical question has gone unanswered: What does the public actually think? We decided to find out.
TLDR: Tens of millions of Americans use prediction markets, Americans want prediction markets regulated but not banned, and Americans’ views of them are still up for grabs.
For this poll, we worked with Echelon Insights to poll 1000 U.S. Likely voters between February 13th and 18th, with the poll entering the field a few days after the Super Bowl to give the electorate time to digest any ads and commentary on these markets around the game. A more detailed methodology summary can be found below.
1) 36% of voters already use prediction markets, and that number should change the entire regulatory conversation. When asked what best described their personal experience with prediction markets, 11% of respondents said they put money on outcomes with them, and 19% said they browse the odds for information but don’t put money down, and 6% say they do both. This is far higher usage than we expected.
2) Usage of prediction markets differs by age. While 38% of those 18-34 and 28% of those 35-49 have put money down via prediction markets, just 3% of those 65 or older have. This represents an age chase even larger than crypto and dwarfs all other demographic divisions. Additionally, persons of color are far more likely to utilize prediction markets than white people, with 68% of white voters saying they never use prediction markets, compared to 53% of black voters and 42% of Hispanic voters. Men are also more likely to use prediction markets, with 46% of men saying they have used them versus 31% of women.
3) Most Americans want prediction markets to be legal but regulated. 35% of respondents felt prediction markets should be legal. A plurality of respondents had mixed views, with 24% saying they should be legal in some instances and 15% saying they should be prohibited in some instances, such as for war and terror contracts. But respondents are more supportive of prediction markets than short selling, which 20% of respondents say should be generally prohibited, but just 29% say should be generally legal.
4) Voters are still formulating opinions on prediction markets. Overall, prediction markets are viewed as 32% favorable and 20% unfavorable among the 95% of respondents who have heard of prediction markets, with 48% stating they have no opinion. There is a big opportunity for prediction markets to tell their story to the American public over the next few months. The claims that voters have decided they dislike these markets are absolutely false, however.
5) The idea that Americans are consistently being bombarded with information about prediction markets is an elite phenomenon. In the last 12 months, only 39% of respondents say they have heard, seen, or read something about prediction markets, while 51% said they have not. This figure is lower than we expected and speaks to the degree that prediction markets are still in their introductory phase to the electorate. Accordingly, just 11% said they were very familiar with prediction markets, with 29% stating they were somewhat familiar, 29% saying they were not very familiar, 20% saying they were not at all familiar, and 11% saying they had never heard of prediction markets.
6) Prediction market users are more socially active than non-users. Prediction market users are substantially more socially active than non-users. 87% of prediction market users talk to their family members weekly, while just 77% of non-users do. 46% of prediction market users play sports or games with others weekly, while just 11% of non-users do. And 19% of prediction market users host people at their house at least once a week, while just 7% of non-users do. Even as society is experiencing fears of reduced social interaction, prediction market users are proving to be a bulwark against those trends. Prediction market users also did not have a substantially different number of close friends than the general population. Overall, 51% people reported having just 1-3 close friends, with just 10% reporting having 8 or more close friends. Of those who personally put money on prediction markets, 53% reported having 1-3 close friends, and 8% reported having 8 or more close friends.
Conclusion
Americans want clear rules that protect participants while preserving the information value these markets provide.
The takeaway for policymakers is clear: Americans aren't asking whether prediction markets should exist; they're already using them. The question now is whether regulation will catch up intelligently, or whether outdated frameworks will push a mainstream financial tool into regulatory limbo. The window to shape this market and public perception of it, is open, but it won't stay open forever.
Methodology
Echelon Insights conducted a survey on behalf of Paradigm to study voter views on prediction markets, fielded online from February 13-18, 2026 in English among a sample of N=1,008 registered voters in the Likely Electorate (LE) nationwide. The sample is a non-probability sample drawn from the Lucid sample exchange based on demographic quotas for registered voters in the likely electorate nationwide, and matched to the L2 voter file to verify respondents’ voter registration status.
Measures taken to ensure data quality included measures to prevent duplicate responses, questions designed to disqualify inattentive respondents, and the removal of respondents from the data file who answered more than one-third of the questions they were asked in less than one-third of the median response time per question.
The sample was weighted to reflect modeled turnout and demographic characteristics of the population of voters in the 2026 likely electorate based on a probabilistic model that incorporates data from the US Census Bureau's American Community Survey and Current Population Survey Voting and Registration Supplement, as well as L2 voter file data. Weighting dimensions included gender, age, race/ethnicity, education, region, party, and voting history.
Calculated the way it would be for a random sample and adjusted to incorporate the effect of weighting, the margin of sampling error is +/- 3.5 percentage point.
Copyright © 2026 Paradigm Operations LP All rights reserved. “Paradigm” is a trademark, and the triangular mobius symbol is a registered trademark of Paradigm Operations LP