This page tracks what prediction markets currently say about the 2026 US midterms. The numbers below are live implied probabilities from Polymarket, refreshed automatically. They are not forecasts and they are not our opinion. They are what traders are collectively willing to pay right now.
What the market is saying
Two things stand out in the current pricing, and they point in different directions.
The House is not being treated as a contest. The market has assigned it lopsidedly to one side, and the price has barely moved in weeks. When a market sits above 85 percent and stops reacting to news, traders have largely stopped arguing about the outcome. That does not make it certain, but it does mean new information is not changing many minds.
The Senate is the opposite. It has been sitting close to a coin flip, which is the most informative state a market can be in. At those levels, a single competitive race breaking one way moves the whole number. This is where it is worth watching the price rather than the headlines.
Underneath the two chamber markets sit the individual races. Most of them are not close. A handful are, and those are the ones the table above sorts to the top: the races where the market itself admits it does not know.
How to read these numbers
A price of 49.5 percent does not mean “the Democrats will narrowly win the Senate.” It means that at this moment, buying and selling has settled at a level where a contract paying out one dollar if they win trades for about 49.5 cents. It is a price, set by people risking money, and it updates continuously.
Three habits make these numbers useful rather than misleading:
- Watch the move, not the level. A market drifting from 42 to 49 over a week is telling you something. A market sitting at 49 is telling you almost nothing except that it is uncertain.
- Check the volume before you trust the price. A race with a few thousand dollars traded can be pushed several points by one order. A race with a million dollars behind it is far harder to move. The volume column above is there for exactly this reason.
- Remember what the contract actually resolves on. Every market has written resolution rules, and they can differ from the plain-English question. Read them before assuming you know what you are buying.
If any of that is unfamiliar, we wrote a longer explainer on how to read prediction market prices, and a broader one on how prediction markets work.
Which races actually matter
For control of the Senate, only a small number of seats are genuinely in play. The table above surfaces them automatically by sorting on how far each race sits from a coin flip. A race at 49 to 48 carries far more information than a race at 95 to 5, because the market is still processing it.
Governor races are included for a simple reason: they attract separate liquidity and they often move on state-level news that the national markets ignore for a day or two. That gap is where the market is least efficient.
Where these markets trade
Two venues dominate US election markets, and they are structurally different.
Kalshi is a CFTC-regulated exchange based in the United States. Its contracts are regulated financial instruments, which matters for anyone who cares about oversight, tax reporting and dispute resolution.
Polymarket is built on public blockchains and typically carries deeper books on political questions, which is why the data on this page comes from there. Deeper liquidity generally means a more informative price.
The two often disagree by a point or two on the same question. That spread is itself information, and it is one of the few genuine edges available to a careful observer. Our full comparison covers fees, resolution and access in detail.
Is trading election markets legal in the US
This is the question most people arrive with, and the honest answer is that it depends on the venue and on where you live. CFTC-regulated event contracts occupy a different legal category from sports betting, and the rules have moved considerably in the last two years. We keep a current breakdown in are prediction markets legal in the US, and a separate piece on whether Kalshi is legitimate and regulated.
How accurate have prediction markets been
Better than pundits, worse than their reputation. Markets priced several past election outcomes more accurately than polling averages, and they also missed badly at times, usually when liquidity was thin or when the question was poorly specified.
The useful framing is not “are they right” but “what are they pricing in that I am not.” A market at 87 percent has already absorbed the obvious arguments. The value is in noticing when it has not absorbed something.
Questions people ask
Are these odds a prediction of who will win?
No. They are prices that reflect what traders are willing to pay, expressed as an implied probability. A 70 percent market loses roughly three times in ten.
How often does this page update?
The live data refreshes automatically while the page is open, and pulls directly from Polymarket.
Why do Kalshi and Polymarket show different numbers?
Different user bases, different liquidity and different resolution rules. Small persistent gaps between the two are normal.
What moves these markets fastest?
Candidate withdrawals, court rulings on maps and eligibility, and unexpected polling in the handful of genuinely close races. National news usually moves them less than people expect.
Bellwether earns a referral commission if you open a market through the links on this page. It does not change which markets we show or how we describe them, and the data is pulled unedited from the exchange. Trading event contracts carries risk of loss.