Political Betting Statistics: $1.5 Billion on the Midterms, a Small Crowd Behind It

Betting on US elections has grown fast. The Anti-Corruption Data Collective counted $123.8 million bet on congressional races across the whole 2024 cycle. By October 29, 2026, betting on 2026 congressional races had reached $431 million, more than three times that. Kalshi alone reported $1.51 billion across its midterms markets by November 3. But the total depends on who’s counting, and much of the money comes from a small group of bettors.

More than $1.5 billion has been bet on the 2026 midterms on Kalshi alone. Counting only congressional races across three platforms, the Anti-Corruption Data Collective put the total at $431 million by October 29. On Polymarket, the top 1% of wallets account for 75% of congressional betting volume.

Figures are as of November 3, 2026, unless noted otherwise.

At a glance

• Kalshi’s Midterms Hub showed $1,510,533,498 in volume on November 3, 2026.
• $431 million had been bet on 2026 congressional races by October 29, 2026, more than three times the $123.8 million bet across the whole 2024 cycle.
• There were 28,845 markets in the 2026 congressional races, 61 times as many as for 2024 congressional races.
• The top 1% of Polymarket wallets held 75% of 2026 congressional betting volume in October, up from 68% in August.
• Two-thirds of congressional markets with live order books could be moved five cents for under $100.
• Kalshi and Polymarket handle about 95% of prediction market volume.
• More than $25 billion was traded across CFTC-registered prediction markets in 2025.
• Only 19% of registered voters say election prediction markets are accurate, compared with 37% for polls.
• In April 2026, Kalshi fined and suspended three congressional candidates for betting on their own races.
• On April 30, 2026, the Senate barred senators, Senate officers and Senate employees from trading on prediction markets.

How Much Has Been Bet on the 2026 Midterms Depends on Who’s Counting

There’s no single figure for how much has been bet on the 2026 midterms. Each source counts a different set of markets, over a different period, in a different way.

The gap comes from what each source includes. The Anti-Corruption Data Collective counts only congressional markets across Kalshi, Polymarket and Polymarket US. Its total leaves out governor and mayoral races, state-level races and referendums, which the platforms also offer. Kalshi’s figure is the platform’s own reported total for its Midterms Hub.

Kalshi dominates the congressional total too. Of the $431 million, Kalshi accounted for more than $286 million, nearly twice the volume of Polymarket and Polymarket US combined.

28,845 Markets on Congressional Races

The midterms didn’t just draw more money. They drew far more ways to bet. By October 29, 2026, Kalshi, Polymarket and Polymarket US had opened 28,845 markets on 2026 House and Senate races, 61 times as many as for 2024 congressional races.

The number of new markets has roughly doubled every quarter. In the third quarter of 2026 alone, 15,876 congressional markets opened. Every congressional race now has trades on at least one platform, and Kalshi offers markets on every race. That works out to an average of 154 markets per Senate seat and 46 per House seat.

Polymarket’s own midterms page shows where the markets cluster. Of its roughly 2,000 midterm markets, 874 are on House races, 569 on margins of victory, 116 on Senate races and 108 on governor races. The Council on Foreign Relations put the total at “more than one thousand” across Kalshi and Polymarket.

A Handful of Bettors Move the Odds

Betting on the midterms is big, but it’s concentrated in very few hands. On Polymarket, the top 1% of wallets held 68% of 2026 congressional betting volume in August. That rose to 70% in September and 75% in October. In the 2024 cycle, it was 79%. The bottom half of wallets accounted for just 0.2% of bets, down from 0.4% two months earlier.

The crowd behind most markets is small. Polymarket has 89,453 unique wallets trading on congressional markets, spread across 7,397 markets. A typical market has about 48 bettors, and only 32 markets have more than 1,000. In October, 84% of Polymarket’s congressional markets had fewer than 100 wallets, up from 80% in August.

That makes many markets cheap to move. 90% of congressional markets have less than $10,000 in volume. Among the 15,985 markets with live order books, two-thirds could be pushed five cents for under $100, and 96% for $3,500 or less, the maximum an individual can donate to a campaign. Even the hardest-to-move market in each race could be pushed five cents for less than $25,000 in 90% of races.

The money piles into a few races. The Texas Senate race, the most bet-on congressional market, neared $64 million. The top five states hold 59% of total volume, down from 74% two months earlier. On Kalshi, 88% of volume is concentrated in just 50 of the 470 races.

Betting isn’t limited to Congress. On November 3, Kalshi’s market on the Los Angeles mayoral race alone showed $109 million in volume.

The wallet figures cover Polymarket only. Kalshi and Polymarket US don’t make individual trader identities public, so their concentration can’t be measured the same way.

Candidates Have Already Been Caught Betting on Themselves

In April 2026, Kalshi fined and suspended three congressional candidates for betting on their own races. They were a Democratic state senator running for a US House seat in Minnesota, an independent running for US Senate in Virginia, and a Republican who lost a House primary in Texas.

The bets were small. One candidate bet $50 that he would win his primary. Another bet less than $100. The fines ranged from $539 to more than $6,200, and all three were suspended from Kalshi for five years. Two of the cases ended in settlements. The third, the largest fine at $6,229.30, was a disciplinary action after the candidate refused to settle and stopped responding.

These weren’t the first cases. In February 2026, Kalshi suspended a California governor candidate who had bet $200 on his own race.

According to the Council on Foreign Relations, Kalshi opened more than 200 investigations in the first half of 2026 and blocked more than 100 potential insider trades in the first quarter. It has also referred more than 50 suspected insider traders to the CFTC, leading to at least 20 investigations.

Betting on your own race isn’t always illegal, though. US commodities law has no general ban on insider trading. A campaign aide who trades on internal polling likely breaks the law, but a candidate who does the same may not.

In the UK, the Gambling Commission charged 15 people with cheating under the Gambling Act 2005 in April 2025. The charges concern bets on the timing of the 2024 general election, and the investigation focused on advance knowledge of the proposed election date.

A $1.25 Million Price Spike Lasted Nine Seconds

In the final two weeks of the 2012 presidential race, a single trader on the exchange Intrade accounted for a third of all money staked on Mitt Romney. On Election Day, his orders held Romney’s odds steady for hours, even as prices on other exchanges moved sharply toward Barack Obama. When he pulled his orders as exit polls came out, Intrade’s price quickly matched other markets, and he lost close to $7 million.

On May 20, 2026, a trader put $1.25 million into Kalshi contracts on one Los Angeles mayoral candidate. The price jumped 16 cents. Within nine seconds, other traders sold into the spike and returned the price to its previous level. The trader lost $1.25 million.

Researchers have tested this on purpose. Economists Paul Rhode and Koleman Strumpf placed random trades in the Iowa Electronic Markets’ 2000 presidential contracts, and prices drifted back toward their earlier levels within a day.

Large bets don’t always lose. In 2024, a French trader who commissioned his own private polls and used 11 accounts on Polymarket made $85 million in profit on the presidential race. Unlike the attempts above, his bets helped push up Polymarket’s Trump price from early October until a few days before the election.

Markets Beat Models on the Big Race, Not the Small Ones

Prediction markets did better than forecasting models on the 2024 presidential race, but not on everything. A study presented at the ACM Collective Intelligence Conference compared Polymarket with three statistical models (FiveThirtyEight, The Economist and Silver Bulletin) across 41 outcomes in the 2024 election.

On the eve of the election, all three models had the presidential race at about even. Kalshi gave Trump a 55% chance, and Polymarket gave him 59%. On the presidential winner, no model beat the market. A trader who bet based on each model’s forecast would have lost between 14.6% and 16.0%.

Lower down the ballot, the results flipped. In 13 competitive House races, both models that published House forecasts beat the market, with returns of 47.4% and 34.2%. The authors point to low trading volume as a possible reason. The House markets drew $2.26 million in total, compared with $3.69 billion on the presidential winner alone.

Across many races, market prices have held up well. Kalshi studied 2.2 million resolved markets, including about 6,000 election markets. It found that when its markets give a candidate a 60% chance, that candidate wins almost exactly 60% of the time. The Council on Foreign Relations reports that the results were independently verified. Economist Eric Zitzewitz studied election markets from 1880 to 2025 and found no systematic bias toward either party, ruling out mispricing of about 5 percentage points or more.

Individual races can still go the other way. In Brazil’s October 25, 2026 presidential runoff, markets gave Luiz Inácio Lula da Silva a 20% chance. He won by more than 4.4 million votes.

Voters Doubt Prediction Markets but Still Follow Them

Most voters doubt prediction markets, especially next to traditional polls. A Brennan Center-funded survey asked 1,994 registered voters on September 21, 2026, whether each is accurate and whether each represents public opinion.

The odds still shape what voters expect. Voters chose between pairs of hypothetical candidates, shown alongside both poll numbers and Kalshi odds that varied at random. The table shows how much more or less likely voters were to pick a candidate as the winner, compared with a candidate at 50%.

Kalshi odds swayed voters almost as much as polls did, and more than any of the candidates’ background traits, including demographics and past experience. Warning voters about prediction markets made no difference.

Those odds now reach far beyond bettors, as the wider prediction market statistics show. CBS, CNN, Fox News, the Wall Street Journal and X have struck deals to show Kalshi or Polymarket odds in their 2026 midterm coverage. A Paradigm/Echelon Insights poll found that more than a third of likely voters have used prediction markets in some form, whether to trade or just to check prices.

Most voters want more regulation. In the Brennan Center survey, 48% said prediction markets should be regulated more strongly than they are now, 23% said some types should be banned, and 16% said all of them should be banned. Only 14% said they shouldn’t be regulated at all. In a Reuters/Ipsos poll, 69% of Democrats and 61% of Republicans opposed using prediction markets to bet on election results.

About 120 Lawsuits Across 28 States

Whether Americans can legally bet on elections is still being fought in court.

Some 120 lawsuits across 28 states involve prediction markets, the states and the Commodity Futures Trading Commission (CFTC), the federal agency that regulates them. The CFTC and the Justice Department have sued nine states to stop them from enforcing their gambling laws against prediction markets.

The federal position has changed quickly. In September 2023, the CFTC blocked Kalshi’s contracts on which party would control Congress, finding they involved gaming. A federal court overturned that order in September 2024, and the CFTC dropped its appeal in May 2025. In February 2026, it withdrew a 2024 proposal that would have banned election contracts. Its June 2026 proposal states that political elections are not “gaming,” which would generally place election contracts outside the rule used to ban them.

Many states disagree. About 40 states have argued in court that prediction platforms offer products nearly identical to gambling. Arizona filed criminal charges against Kalshi over election bets, though a federal judge blocked the prosecution. Minnesota passed a law banning Kalshi and Polymarket this year, and a federal judge blocked it before it took effect. In October, Utah sued Kalshi in state court to stop its election contracts.

Congress has moved more slowly. On April 30, 2026, the Senate adopted S. Res. 708 by unanimous consent, barring senators, Senate officers and Senate employees from trading on prediction markets. The House has no equivalent rule, though two resolutions are pending. More than a dozen measures on prediction markets are pending in Congress overall.

Two states still bar election bettors from voting. Wisconsin does so under a law dating to 1849, and New York under its constitution. In July 2026, Wisconsin’s Elections Commission warned that trading on a race and then voting in it could be a felony.

Election Betting Is Older Than Polling

Americans were betting on elections long before polls existed. In one 1796 congressional race in New York, a politician bet “more pounds than I am worth pence” that his patron would win by a thousand votes. The patron lost by 18 ballots.

By the late 1800s, election betting was big business. From 1868 to 1940, organized markets in New York and other cities took bets on presidential races, and New York handled more than half of all election betting. At times, betting on elections at New York’s Curb Exchange outpaced trading in stocks and bonds.

The peak came in 1916, when New York bettors wagered $10 million, about $300 million in 2026 dollars. That was more than twice what the presidential campaigns spent that year. Betting was also a social ritual. In the 1900 election, there were an estimated half a million “freak bets,” in which losers ate crow or pushed winners around in wheelbarrows. That’s about one bet for every 30 voters.

The markets were good at picking winners. In the 15 presidential elections from 1884 to 1940, the mid-October betting favorite won 11 times. The underdog won only once, in 1916.

Then the markets faded. Scientific polls took over after George Gallup correctly called the 1936 election, and New York legalized betting on horse races in 1939, which drew gamblers away. By the 1944 presidential election, betting commissioners were hard to find on Wall Street.

Election betting came back in smaller forms, mostly with tight limits:

$173 Million Bet on Canada’s Elections, Where Both Platforms Are Barred

Prediction markets now take bets on elections around the world, including in countries that don’t allow them. The table shows the five countries outside the US with the most traded election markets, by the platforms’ own stated volumes, and whether each platform bars that country’s residents from trading.

Canada bars residents from both platforms, yet its elections still drew $173 million. In January 2026, Portugal banned Polymarket after a €4 million surge in betting on its presidential election in the final hours before results were known.

Some of this volume isn’t real betting at all. Wired reported that Ethiopia’s presidential election saw $284 million traded on losing candidates on Polymarket, all of it between pairs of accounts that made no real gain or loss. In 2026 US congressional markets on Polymarket, the Anti-Corruption Data Collective found 591 suspicious round-trip trades, in which wallets bought near-certain contracts and sold them back within six hours. $618,395 moved through these trades, for a total profit of $34.

What Nobody Publishes

Some of the most important numbers about election betting aren’t public. None of the sources behind this article publish them.

The biggest gap is who is behind the money. The Brennan Center notes that the money behind prediction markets isn’t disclosed, so voters can’t see who is shaping the odds in a given race. The Council on Foreign Relations recommends a public disclosure rule for large positions, but none exists yet. Offshore platforms add another layer. Polymarket’s international platform isn’t registered with the CFTC, and many offshore platforms let traders stay anonymous.

There’s also no single agreed total. Each platform reports its own volume, the Anti-Corruption Data Collective recalculates it with its own method, and some published totals give no method at all.

Nobody publishes how much ordinary bettors lose either. The Council on Foreign Relations says most people who bet on election markets lose money over time, and recommends that the CFTC study trading data to measure it. No public figure exists yet.

Ultimately, full totals for the 2026 midterms aren’t available yet. Betting continues through Election Day and until markets settle, so final figures will only come after the results.

How to Check These Numbers Yourself

Most figures in this article can be checked in a few clicks. Live totals change by the minute, so expect them to be higher than the figures above.

• Kalshi’s midterm volume ($1.51 billion). Open Kalshi’s homepage and find the “Midterms Hub” card. The total volume is listed under the title.
• Congressional betting totals ($431 million, 28,845 markets, 75%). Open the Anti-Corruption Data Collective’s Briefing 4. All three figures are on the cover page.
• The revised 2024 total ($123.8 million). In the same briefing, see Figure 1 on page 3. A footnote on that page explains why ACDC raised its earlier $92 million count.
• How cheaply markets can be moved. Briefing 4, Figure 5 on page 6, shows how many markets fall into each cost range.
• Polymarket’s market counts. Open Polymarket’s midterms page. The number of markets in each category (House, Senate, governor and so on) appears next to each tab.
• Voter survey results. The Brennan Center’s article shows the full survey charts, including the agree and disagree figures for prediction markets and polls.
• The Senate trading ban. Search “S. Res. 708” on Congress.gov. The page shows it was agreed to on April 30, 2026.

Sources

Every figure above traces to one of these.

1. Anti-Corruption Data Collective, Gambling on Democracy, Briefing 4: A Midterms Update and the Global Pictureacdatacollective.org
2. Council on Foreign Relations, Betting on Democracy: The Promise and Perils of Election Marketscfr.org
3. Kalshi, Midterms Hubkalshi.com
4. Polymarket, Midtermspolymarket.com
5. ACM Collective Intelligence Conference, Political Prediction and the Wisdom of Crowdsdl.acm.org
6. Federal Register, Prediction Markets; Public Interest Determinationsfederalregister.gov
7. Mayer Brown, The Odds Are In: CFTC Proposes Framework for Event Contracts and Prediction Marketsmayerbrown.com
8. Congress.gov, S. Res. 708congress.gov
9. NBC News, Kalshi fines and suspends three politicians for ‘insider trading’ on their own racesnbcnews.com
10. CNN, Kalshi prediction site suspends three political candidates for betting on their own racesus.cnn.com
11. UK Gambling Commission, Gambling Commission charges 15 with General Election betting offencesgamblingcommission.gov.uk
12. Brennan Center for Justice, New Study Finds Prediction Markets Are Influencing Voters, and Voters Want Them Regulatedbrennancenter.org
13. Journal of Economic Perspectives (Rhode & Strumpf), Historical Presidential Betting Marketsideas.repec.org