Prediction Markets Face Their Biggest Test Before Midterms

Prediction markets sell themselves as the most honest election forecasters ever built, so I'll start with the counter-case. Midterm election betting is heading into a stretch where a federal regulator is reportedly examining Polymarket, that same regulator is suing a state, and much of the reassurance about clean markets comes from the venues themselves.

I've called this their biggest stress test yet, and I'll own that it's my opinion, not a finding. Nothing in the reporting establishes that Polymarket broke any rule, and the court losses so far are about sports contracts. But two stories landed on October 27, the Polymarket CFTC probe and the CFTC's lawsuit against Utah, and they're easy to confuse.

I care less about who wins on November 3 than whether these platforms survive the rules fight. Below: what's reported, with dates and sources, how to read the odds without fooling yourself, and where the real risks sit. No picks.

How do prediction markets work, and how do you read the odds?

A prediction market lists a question about a future event and sells contracts on each outcome. Prices are quoted in cents, and a winning contract pays $1. Kalshi describes itself as an event-contract exchange subject to CFTC oversight, and on its Congress balance-of-power page a 65¢ price implies roughly a 65% chance.

Here's my first skeptical point: a price is a live clearing level, not a forecast. Kalshi's midterms page says the odds are live prices that change as traders react to polls, fundraising and news. Nobody is running a model. People are putting money behind opinions, and the number is simply where buyers and sellers met.

Details that trip people up:

• The two sides don't add to exactly 100. On October 27, Kalshi showed House Democrats at 90.6% and Republicans at 9.6%, and the Senate at 64% and 37%, per Kalshi's own blog. The gap reflects the bid/ask spread.
• A 91% price still leaves a 9% chance the other side wins.
• Venues can disagree on the same event, and contract wording can differ, so read each venue's rules page before comparing prices.

What is the CFTC reportedly probing at Polymarket?

Start with how thin the sourcing is. Reuters first reported on October 27 that the CFTC is examining whether Polymarket violated its 2022 settlement, and CNBC, citing a person familiar with the matter, says that is the question. All of it is anonymous. Polymarket declined to comment, the CFTC did not respond, and no start date was given.

The 2022 deal is the backdrop: Polymarket stopped U.S. users from accessing its platform and paid a $1.4 million civil penalty. According to CNBC, the new inquiry looks at:

• Polymarket's controls for detecting where customers are located
• Its U.S. advertising
• The U.S. platform that launched in late December, specifically customer onboarding and protection measures

CNBC's source says the inquiry concerns possible past breaches and is "unrelated to the platform's present operations." It is not about the midterm bets, and coverage describes an investigation, not a finding or a charge.

It also sits inside a broader probe the CFTC opened earlier this year, after a Wall Street Journal exposé said Polymarket paid content creators to post videos of simulated trades and fabricated wins on dummy versions of its site. If that allegation holds up, it's the more uncomfortable thread, because it's about how the product was sold.

Which midterm bets did Polymarket reportedly refer to the CFTC?

This comes from NPR's reporting, relayed by Decrypt on October 27. Shana Bautista, Polymarket's head of global intelligence and investigations, told NPR the company has investigated "dozens" of bets in its midterm election markets. At least two were referred to the CFTC, NPR reported, citing a person familiar with the probes. Polymarket hasn't confirmed the referrals on the record in that coverage, so treat them as reported, not established.

The monitoring system pulls data from social media, FEC filings, crypto companies, banking partners and statistical models. It alerts Bautista when a user trips enough internal red flags, and Decrypt says it generates hundreds of alerts a day.

Here's how I read that. A company reporting dozens of investigations and hundreds of daily alerts is showing you effort, not outcomes. Alerts aren't violations. It's good that the surveillance exists; the harder question is whether it catches the bets that matter before the market moves, and nobody outside the company can check that.

Keep this separate from the 2022-settlement probe above. One is about suspicious trading, the other about who can access the platform.

Why is the CFTC suing Utah over prediction market regulation?

Because Utah went after Kalshi, and the CFTC says that's its turf. On October 20, Cointelegraph reported that Utah Attorney General Derek Brown sued Kalshi, its affiliates and two Susquehanna trading companies, alleging illegal betting in the state since 2024. Utah says Kalshi served more than 32,000 users and wants roughly $30 million to $60 million in profits returned, with the exact amount to be set in discovery.

Brown's ask goes beyond sports. He wants the court to stop "sports, elections, politics, culture and mention bets" in Utah.

The Block reports that the CFTC then sued Utah and Brown in the U.S. District Court for the District of Utah, Central Division. It's the agency's tenth suit against a state, after Arizona, Connecticut, Illinois, New York, Wisconsin, Minnesota, Rhode Island, New Mexico and Kentucky. Chair Michael Selig's argument: "Congress entrusted the CFTC to be the federal regulator with sole responsibility to oversee these markets."

Two points of fairness. Preemption is a serious legal argument, not a stunt, and Selig also says event contracts give the public information about outcomes. And Utah's target is Kalshi, a CFTC-registered exchange, not Polymarket. Lumping the two fights together is how bad takes get written.

Can states regulate prediction markets? Where the courts stand

The scorecard so far leans toward the states, but only on sports. On September 25, CNBC reported that a unanimous Sixth Circuit panel said Ohio and Tennessee may apply their gambling laws to Kalshi's sports event contracts. CNBC adds that it's widely expected the question of who regulates prediction markets' sports contracts will reach the Supreme Court.

Election contracts are a different animal, and Utah's request explicitly reaches them. That's why this matters for November even though the rulings I'm citing concern sports.

The platforms' counter-case is real, though. The CFTC says Congress made it the sole regulator, and the dissent in the Third Circuit's Kalshi opinion notes that the CFTC issued an advance notice of proposed rulemaking on prediction markets and that Rule 40.11(a)(1)'s ban on gaming contracts remains in place. It's a dissent, so I'm using it only to show the rulebook is unsettled, not empty. A rule banning gaming contracts is also exactly what critics will quote back at the agency.

What that means in practice:

• Your access can change depending on your state and the type of contract.
• A venue's loss in one court doesn't dictate the result in another.

What do midterm election betting odds say right now?

All 435 House seats and 35 Senate seats are up on November 3. As of October 27, Kalshi's midterms page showed Democrats at 65% to control the Senate and 91% to control the House. Its Congress balance-of-power market at 8:37 p.m. ET that evening showed:

• Democrats sweep: 63%
• Democratic House, Republican Senate: 29%
• Republican sweep: 9%
• Republican House, Democratic Senate: 1%

Those add up to 102, the spread again. Kalshi's own blog says its traders forecast an average of 238.3 Democratic House seats, up from 226 on May 15. That's Kalshi's marketing, so weigh it accordingly. In Texas, the Senate race showed Talarico at 66% and Paxton at 35%, a total of 101.

My skeptical read: CNBC's October 27 report says polls show Democrats leading for the House while the Senate is neck-and-neck. A 65% Senate price sounds more confident than neck-and-neck, and prices fold in money, fundraising and news, not just polls. The direction matches; the confidence is the part to question.

Venues also drift. CNBC reported on September 16 that Kalshi gave Democrats a 55% Senate chance and Polymarket 59%, a four-point gap on the same event. These are crowd probabilities as of those dates, and they keep moving.

What are the biggest risks of trading on prediction markets?

Here are the risks in the order I'd worry about them:

• Thin liquidity. In a market with few traders, one large order can move the price, and a number that moves on one order says little about what the crowd believes. Check volume before reading a probability into any narrow race.
• Whales. A few large accounts can swing odds in small markets. Concentration isn't proof of manipulation, and concentrated markets can still price well, but it's a reason not to hear every number as the voice of thousands.
• Insider bets. Polymarket says it has handled "dozens" of midterm cases, per Decrypt, and its international side allows crypto trades without public identity verification.
• Legal access. Whether you can trade depends on your state and the venue. Utah's suit and the Sixth Circuit ruling show how fast that can change, and Polymarket's 2022 settlement is the reason U.S. access is a live question at all.
• Consumer harm. CNBC notes escalating concern from mental-health experts and lawmakers about how fast these platforms are growing.

Practical takeaways: treat a price as a probability that can be wrong, check volume and the rules page before leaning on it, only risk money you can afford to lose, and confirm what's legal where you live. None of this is advice.

What's still unknown about the Polymarket probe and the Utah fight?

Quite a lot, and I'd rather list it in one place than sprinkle hedges through the piece:

• Whether the CFTC's Polymarket inquiry touches any election market, and where it leads. Nothing so far establishes a violation.
• The identity, size and outcome of the "at least two" cases reportedly referred to the CFTC, including whether the agency acted. No market, wallet or alleged conduct has been made public.
• The exact filing date and relief sought in CFTC v. Utah. The Block puts the filing on Tuesday, October 27; other outlets say October 26. The complaint is the document to check.
• Whether the Supreme Court has agreed to hear any prediction market case.
• Whether any appeals court has ruled on election contracts specifically, as opposed to sports.
• How election contracts settle if control of a chamber is disputed, and on what timeline.
• Whether the Wall Street Journal's allegations about paid creators and fabricated wins lead to any CFTC action.

Until those get answered, anyone who tells you how this ends is guessing.

Bottom line

My read, after working through the counter-case: the stress test is real, but it's mostly a legal-access test, not an accuracy test. On the reporting, the Polymarket probe is a question about past U.S. access controls. The Utah fight is a federal-versus-state jurisdiction fight over Kalshi, and a unanimous Sixth Circuit panel has already sided with states on sports contracts. Neither says the midterm odds are wrong. Both say the ground under these venues is shakier than the price tickers suggest.

So treat a midterm price as a crowd's probability on a given day, not a forecast. Check volume and rules before leaning on it, and remember your state may matter more than the number. I'm not telling you what to bet or whether to bet. I'm telling you which answers would change my mind: what the two referred cases were, what the CFTC's complaint actually asks for, and whether the Supreme Court steps in.