Midterm Prediction Markets Reach $221 Million Ahead of Nov. 3 Vote

Trading in 2026 congressional prediction markets reached $221 million by Sept. 30, surpassing the entire 2024 cycle as thin order books complicate comparisons with election polls.

The total covers 21,811 markets across Kalshi, Polymarket and Polymarket US. Congressional trading reached $92.4 million in the entire 2024 cycle. By Aug. 10, contracts covering 460 of the 470 seats on the Nov. 3 ballot had drawn at least $1 in trading.

That reach does not mean every market has enough activity to produce a stable price. Among congressional markets with live order books on Sept. 30, 8,224, or 55.1%, could be moved 5 cents from their midpoint with less than $1 in notional trading. In those markets, a small trade can change the displayed odds.

Polls and contracts also measure different things. A poll estimates voter preferences from survey responses. A contract price expresses the probability traders assign to the outcome specified in that contract’s rules. Neither a poll result nor a market price is the election result.

Kalshi lists contracts on individual races and control of the House and Senate in its Midterms Hub. The range of contracts creates opportunities to compare forecasts across races, but the amount traded in each market matters when interpreting its odds.

The platforms’ broader growth offers little guidance on that question. Combined global monthly volume on Kalshi and Polymarket rose from about $25.7 billion in May to $53 billion in July. Sports accounted for most of the increase, and those totals cover far more than congressional contracts.

An analysis of more than 72 million Kalshi trades found that contract prices broadly matched how often outcomes occurred across many types of markets. It did not compare 2026 congressional forecasts with polls. Earlier Iowa Electronic Markets data include presidential-election comparisons; prediction markets’ historical edge over polls did not hold in 2024.

NEXTPredict summit host Chuck Todd raised the relationship between the two approaches in a September conversation: “How much do they need each other? How much do they feed each other?” The summit is scheduled for Oct. 22-23 in New York.

The Nov. 3 results will allow forecasts to be checked against outcomes across congressional races. Determining whether markets outperformed polls will require comparing the same contests at the same points in time.