Politics · United States · 4 hrs ago
US midterms test whether prediction markets can beat polls
Prediction markets and polls both favor Democrats to win control of the US House and Senate in the midterms.
Prediction markets let people trade contracts tied to whether events happen, with prices used as estimates of the chances.
Markets are slightly less confident than polling models about Democrats taking both chambers, though the difference has narrowed as Election Day approaches.
On Tuesday, markets put Democrats’ chance of winning the Senate at 77.3%, compared with 80.2% from polling models.
For the House, markets put the chance of Democrats taking control at 94%, while polling models put it at 96.2%.
The fast-growing industry faces greater scrutiny as trading on election outcomes rises and news organizations incorporate market odds into coverage.
A recent Brazil election forecast was badly wrong: markets gave Luiz Inácio Lula da Silva about a one-in-five chance, but he won the presidential runoff.
The US midterms are a major test of whether prediction markets can reliably forecast election results compared with polls.
US midterm elections are a major test of whether prediction markets can forecast results more accurately than polls.
Both prediction markets and poll-based models favor Democrats taking control of the House and Senate, but markets are somewhat less certain.
On Tuesday morning, prediction markets put Democrats’ chance of taking the Senate at 77.3%, compared with 80.2% from polling-based models.
For the House, markets gave Democrats a 94% chance of flipping the chamber, compared with 96.2% from polls.
The scrutiny follows prediction markets giving Brazil’s eventual presidential election winner, Luiz Inácio Lula da Silva, about a one-in-five chance on election day.
- Who
- Prediction-market platforms and poll-based models are forecasting the US midterms.
- What
- The election is testing whether prediction markets can forecast results as well as or better than polls.
- When
- Election Day is Tuesday, 3 November 2026.
- Where
- United States.
- Why
- Prediction markets claim that traders’ financial stakes make their forecasts more accurate, while the midterms have drawn scrutiny after markets gave a low chance to Brazil’s eventual winner.
Prediction markets
Poll-based models
Senate control
Prediction markets
Put Democrats’ chance of taking control at 77.3% on Tuesday morning.
Poll-based models
Put Democrats’ chance at 80.2%.
House control
Prediction markets
Put Democrats’ chance of flipping the chamber at 94%.
Poll-based models
Put Democrats’ chance of flipping the chamber at 96.2%.
Forecasting claims
Prediction markets
Their supporters argue that markets can be more accurate than polls because traders have money on the line.
Poll-based models
Poll-based models provide a separate forecast; the sources do not state a response to the markets’ claims.
Trading volume on Polymarket and Kalshi combined reached just over $4m in the period before the midterm elections.
Trading volume on the two platforms surpassed $90m over the comparable period.
Lula won Brazil’s presidential runoff with 51.8% of the vote after prediction markets had given him roughly a one-in-five chance on election day.
US midterm Election Day brought prediction markets and poll-based models into direct comparison.
- Senate, prediction markets
- 77.3% chance of Democratic control on Tuesday morning
- Senate, polling-based models
- 80.2% chance of Democratic control on Tuesday morning
- House, prediction markets
- 94% chance Democrats would flip the chamber
- House, polling-based models
- 96.2% chance Democrats would flip the chamber
- Kalshi midterm-market volume
- More than $1.2bn traded on the platform
- Brazil runoff result
- Lula won 51.8% to 48.2%
3 sources
What people are saying
help the public forecast, plan for, hedge, and even harness