The upcoming US midterm elections mark a historic shift as the first congressional race where citizens can legally place large-scale bets on the outcome. According to an analysis by Handelsblatt covering 1,930 betting markets across Polymarket and Kalshi, the total trading volume has reached nearly half a billion dollars. This surge represents more than double the amount wagered on individual congressional races during the entire 2024 election cycle, with bets increasing by another third in the final week alone.
Market data indicates a decisive trend favoring the Democratic Party. Of the 52 major price movements observed in the last month, 51 shifted in favor of Democrats. Traders currently assign a 94% probability to Democrats reclaiming the House of Representatives and a 73% chance of them securing the Senate, up from 47% in early September. This sentiment suggests President Donald Trump could lose control of the entire Congress for the second half of his term. In total, approximately $140 million is currently wagered on Republican losses compared to only $61 million on their victories, creating a betting ratio of more than two-to-one against the President's party.
The markets have effectively decided many races in advance. Out of 484 contests analyzed, traders view 390 as having a favorite with at least a 90% chance of winning, leaving only 25 races considered open duels. High liquidity is concentrated in specific battlegrounds, such as Texas, where the Senate race between Ken Paxton and James Talarico has attracted about $79 million in bets. While Paxton was once favored, his win probability on Polymarket has dropped from 48% to 27%, whereas Talarico now holds a 74% chance of victory, which would mark the first time in over 38 years a Democrat wins a Senate seat for Texas.
Other notable markets include Ohio, where Sherrod Brown is favored at 71% against Jon Husted with $12 million wagered, and Maine, where incumbent Susan Collins faces a 27% win probability against Troy Jackson despite winning every election since 1996. In California, the governor's race shows Xavier Becerra as a near-certain winner at 97%, yet the market volume exceeds $76 million. A significant portion of this activity involves betting on longshots; for instance, $27 million has been placed on Steve Hilton, who holds only a 3% chance of winning.
Experts note that these markets often suffer from a "favorite-longshot bias," where extreme underdogs are systematically overpriced. Research indicates that bets involving such longshots lose in 98% of cases since early 2025. Furthermore, profitability is highly concentrated; an analysis by the Wall Street Journal found that while over 70% of Polymarket users lose money, 67% of total profits flow to just 0.1% of accounts, primarily professionals and trading firms using algorithms. Kalshi acknowledges a similar dynamic, noting there are 2.9 losers for every winner.
The predictive accuracy of these platforms remains a subject of debate. While they successfully predicted 85% of governor primary races last year, past performance in Brazil showed discrepancies where markets favored the wrong candidate before an election. Additionally, risks of manipulation exist, with Kalshi already banning six candidates who bet on their own races ahead of the midterms. Despite these concerns, Federal Reserve research suggests these markets can match professional economic forecasts, offering a distribution of probabilities rather than a single survey number. Ultimately, while traders have placed billions on a Democratic sweep, the final result rests with the voters.
Sources: Handelsblatt.