NextFin News - The 2026 US midterm elections are unfolding as the most heavily bet-on congressional contest in history, with more than $1.48 billion wagered across prediction markets - and yet politics now accounts for just 1% of trading volume on the two biggest platforms, Kalshi and Polymarket, down from two-thirds during the 2024 presidential election. The midterm frenzy is not a political story. It is proof that the prediction-market boom has outgrown politics entirely.
While headlines focus on who will control the House and Senate, the real story is what traders are betting on instead. Prediction-market volume reached $188 billion in the third quarter of 2026, up nearly 70% from the previous quarter and 21 times year-earlier levels, with sports - not elections - driving the expansion. The industry that promised to turn democracy into the ultimate forecasting machine has quietly become a sportsbook with a politics section.
The Numbers: A Political Surge That Is Mathematically Invisible
The scale of the disconnect is stark. In October 2026, politics represented just 1% of combined trading volume across Kalshi and Polymarket, according to platform data. During the 2024 presidential election, the same category made up roughly two-thirds of activity. Even with more than $1.48 billion wagered on the 2026 midterms, election trading remains a rounding error in a sector that posted $44 billion in total volume in 2025.
The contrast is even sharper against the sector's peak. Weekly trading volume across prediction platforms exploded from $500 million in June 2025 to nearly $6 billion in January 2026, data from blockchain analytics firm Dune shows, before settling above $4 billion in weekly volume in December 2025. Against that backdrop, the entire midterm cycle - accumulated over months - is about a quarter of a single peak week.
Kalshi reported $1.33 billion across all midterm bets as of Monday, while the Anti-Corruption Data Collective recorded about $145 million on Polymarket's congressional markets. A separate tally of all midterm-related markets, including Polymarket's offshore site, put that platform above $340 million. ForecastEx, the Interactive Brokers-owned venue, added more than $6.4 million. More than 28,000 midterm markets have opened this cycle, but a fraction of individual contests - Senate races in Texas, Michigan, Maine and South Carolina - account for most of the volume.
"It's sizable," said Dustin Gouker, an industry expert who publishes a newsletter on prediction markets. "It's still much less than sports and crypto markets, but it's a lot of trading. And it's valuable. It's a meaningful signal that provides probabilities that aggregate polls and public sentiment."
One caveat matters for reading these figures: on Kalshi, each contract is counted at its $1 notional value rather than the price paid. Buyers paid about $41 billion for $142 billion in contracts last quarter, meaning headline volume overstates the actual cash at risk by roughly a factor of three to four.
Why Politics Lost Its Crown
Three forces explain the shift, and none of them are about election accuracy.
First, sports is simply bigger and stickier. Sports represented around 40% of Kalshi's quarterly volume in the third quarter of 2026, according to data analytics firm CryptoRank - and that figure refers to Kalshi alone, not the sector. Sports markets run year-round, settle quickly, and generate constant trading flow. Elections are episodic: a burst of attention followed by months of thin volume. For a platform chasing a $188 billion quarterly run rate, sports is the reliable engine; politics is the seasonal decoration.
Second, the regulatory path for sports and crypto contracts is cleaner than for election markets. In May, the Commodity Futures Trading Commission filed a brief in the Sixth Circuit asserting exclusive federal authority over prediction markets and arguing that federal derivatives regulation preempts state gambling laws applied to CFTC-regulated exchanges. On September 22, CFTC staff issued guidance distinguishing "mention markets" - contracts on whether a named person says particular words, attends an event, or interacts with someone - from outcomes generated through broader processes like elections or economic data releases. The guidance does not create new obligations, but it signals where regulators see the sharpest manipulation risk: contracts whose outcomes a single person can influence. Senator Elizabeth Warren and Republican Senator John Curtis pressed the same concern in an October 28 letter to Kalshi and Polymarket, raising questions about election-related event contracts and possible market manipulation.
Third, the 2024 election set an impossible comparison. Prediction markets rose to prominence by correctly signaling Donald Trump's victory when poll-based models called it a tossup. The day before that election, prediction markets gave Trump between a 57% and 62% chance of victory while many polls showed a dead heat. That one hit created an expectation that prediction markets would consistently beat polls. They have not. Markets implied roughly a 20% chance that Brazilian President Luiz Inacio Lula da Silva would defeat right-wing rival Flavio Bolsonaro last month - and Lula won, a high-profile miss that left the markets on the wrong side of the upset.
The gap between markets and polls has persisted through the 2026 cycle, though it has narrowed. For the Senate, the difference between prediction-market odds and polling-based models averaged about 7.4 percentage points since September 1, sitting at 7.2 points on one recent Monday with three prediction markets placing a 73.1% chance of Democrats taking control versus 80.3% from two polling models. Around the House, the average gap has been less than a percentage point in recent months, narrowing to 2.6 points on the latest reading: 94.1% from prediction markets versus 96.7% from polls. Prediction markets have been the less confident of the two forecasters - the opposite of the 2024 narrative.
The Concentration Problem: Thin Markets, Loud Prices
Beyond volume share, the structure of political prediction markets raises questions about how much weight their prices deserve. On Polymarket's global platform, the top 1% of wallets account for 68% of trading volume on 2026 congressional markets, and 80% of those markets have fewer than 100 wallets trading, according to an August analysis by the Anti-Corruption Data Collective, a Washington nonprofit. The group analyzed more than 21,000 markets on 2026 US congressional midterm elections across Kalshi and Polymarket's international and US platforms through September 30, comparing them with 463 markets on the 2024 congressional races.
"Many of these contests are close," said Yesha Yadav, associate dean at Vanderbilt University Law School. "Tuesday does, in fact, represent a real test for the new wave of prediction markets."
The concentration cuts both ways. A small proportion of sophisticated profit-seeking traders appears to account for a large share of capital in prediction markets, according to Harry Crane, a statistics professor at Rutgers University, and to ACDC data. Those traders, using polling and other data, help correct prices - which is why the markets should be taken seriously, Crane said. But it also means a handful of well-funded accounts can move prices in races where fewer than 100 wallets are active.
Platforms say they have ramped up controls. Kalshi blocks politicians and campaign workers from political wagers and has already suspended six 2026 candidates for betting on their own races. The companies also say their markets correct quickly when large bets move prices.
"The point is not that prediction markets are perfect, it's that in an imperfect world full of chaos and biased news sources, it's important to have another way to forecast important events," said Sean Patrick Maloney, CEO of the Coalition for Prediction Markets.
What the Election Will Actually Prove
Academic research generally finds prediction markets well calibrated at scale. A 2025 Vanderbilt University paper analyzing nearly 2,500 markets on the 2024 election found that contracts priced at a given probability tend to resolve at roughly that rate across large samples. But calibration is not the same as accuracy on any single race.
"But how accurate the price is for any specific market is impossible to know," said Joshua Clinton, a political-science professor at Vanderbilt who led the 2025 research. Most races are not competitive, which inflates headline accuracy while leaving the interesting contests - the ones traders actually care about - much harder to call.
On Monday, prediction-market prices aggregated by analytics firm BetInt implied Democrats had a 95% chance of taking the House of Representatives and a 76% chance of taking the Senate. Both Kalshi and Polymarket separately assigned roughly 95% odds to Democrats taking the House, with Senate odds above 75%. If those odds hold, the industry scores a win. If Democrats fall short in either chamber, the reputational damage could land harder than a normal forecasting miss - because this time the markets are being watched by everyone from campaign war rooms to television news desks.
The Second-Order Read: Prediction Markets Are Becoming Something Else
The deeper implication is structural. The industry's original promise - that betting markets would become the definitive source of political truth - has given way to a different business: event contracts as a general-purpose trading venue. The 2024 election was the breakout moment, but the 2026 midterms show the breakout was not about politics. It was about the format.
This is a structural shift, not a cyclical dip. Three pieces of evidence support that call. First, the volume composition has permanently reweighted: sports and crypto now dominate a revenue base that grew 21-fold year over year, and there is no mechanism that would push politics back to two-thirds of volume. Second, the regulatory framework is crystallizing around non-political products - the CFTC's May brief and September guidance carve out clearer lanes for sports and event contracts while flagging manipulation risk in person-specific political markets. Third, the competitive landscape is shifting toward general event contracts: CME Group and FanDuel announced a partnership to bring event contracts to mainstream audiences, covering the S&P 500, oil, gold, and crypto rather than elections.
The cyclical counterpoint matters too. Election volume is inherently lumpy - it spikes around debates, primaries, and Election Day, then fades. A strong final-day surge in midterm betting could make the political share look healthier in November's numbers. But even a doubling of election volume would leave politics in the low single digits of a sector trading hundreds of billions per quarter.
The Counter-Thesis: Maybe Politics Is Still the Point
The strongest argument against this reading is that election markets were never meant to be judged on volume share. Their value is informational, not financial - a single well-capitalized market can produce a better probability estimate than a room full of pollsters, regardless of how much money changes hands. Sean Patrick Maloney's Coalition for Prediction Markets makes exactly this case: the point is having another forecasting channel in a chaotic information environment.
That argument holds for the markets that matter. But it does not survive the concentration data. When 80% of congressional markets have fewer than 100 wallets trading, and the top 1% of wallets control 68% of volume, the "wisdom of crowds" is really the wisdom of a few dozen accounts - some of them candidates' own campaigns, before suspensions. A thin market can still be right, but its price is fragile, and its claim to outperform polls rests on the same assumption that failed in Brazil: that sophisticated traders have correctly interpreted the information.
The falsifying signal for the structural-shift thesis is specific: if politics returns to 40% or more of combined Kalshi-Polymarket volume for two consecutive quarters after the 2026 cycle closes - and stays there through the 2028 presidential primary season - then politics remains the core product and the sports surge was the distraction. That would require election volume to grow far faster than sports and crypto, reversing a reweighting that has already proven durable for more than a year.
What to Watch
In the short term, watch the resolution of the Senate races in Texas, Michigan, Maine, and South Carolina - a fraction of the more than 28,000 midterm markets opened this cycle, but the ones accounting for most trading volume. A Democratic sweep would validate the 76% implied probability; a split would hand critics a high-profile miss.
In the medium term, watch the CFTC's final rules on prediction markets, part of a package the administration is finalizing, and whether the jurisdictional fight between federal and state regulators resolves in favor of the CFTC's preemption argument. That decision will determine which products can scale.
In the long term, watch whether the entrants - Coinbase, Crypto.com, traditional sports-betting firms, and the CME-FanDuel venture - treat politics as a loss leader for customer acquisition or avoid it altogether. Their product roadmaps will reveal what the industry actually is.
The Bottom Line
The 2026 midterms are a stress test for prediction markets' accuracy, but they are also a tombstone for prediction markets as a political project. The boom is real - $188 billion in a single quarter, 21 times the year-earlier level - but it is not a political boom. Politics went from two-thirds of volume to 1% in two years, and nothing in the structure of the industry suggests it is coming back.
The markets that promised to forecast democracy have instead financialized everything except democracy. That is not a failure of the model. It is the model finding its real market.
Explore more exclusive insights at nextfin.ai.
