How do prediction markets work? Polymarket, Kalshi and the fight over event contracts

Who this is for: Institutional investors, founders, policy staff and serious retail traders who want to read a prediction market price correctly, know where the regulatory lines sit in late 2026, and price a position of their own including fees and spread.

Prediction markets went from academic curiosity to one of the fastest growing corners of finance in under two years. In October 2024 the open question was whether Americans could legally trade on elections. By mid 2026 the owner of the New York Stock Exchange had committed up to $2bn to Polymarket, Kalshi had raised at a $22bn valuation, and the two platforms were clearing more than $50bn of contracts a month, most of it on sports.

Growth has outrun understanding. A prediction market price is not a poll, not a forecast and not quite a bet. It is the clearing price of a binary derivative, shaped by fees, spreads, liquidity, resolution rules and the incentives of whoever is quoting it. This guide covers the mechanics, the two leading venues, the volume boom, the court fight between the CFTC and the states, accuracy research, how institutions and media use the data (including Crypto Coin Show’s own price predictions page, which runs on live Kalshi odds), the Polymarket token question, and a worked $10,000 example.

Prediction markets by the numbers

What an event contract is and how it is priced

The unit of every prediction market is a binary event contract. It names an event, a resolution source and a deadline: “Will Bitcoin close above $150,000 on December 31, 2026, per the exchange’s settlement index?” It has two sides, Yes and No. At resolution the correct side pays $1 (1 USDC on Polymarket) and the wrong side pays nothing. Before resolution both sides trade freely, and because exactly one will be worth $1, Yes plus No should always sit close to $1.

That is why a price reads as a probability. Buying Yes at $0.43 risks 43 cents to win 57 cents, a fair trade only if the event happens 43% of the time. The read is approximate: fees, the spread, time value on long-dated contracts and the tendency to overpay for long shots all push the raw price away from the true probability.

Order books versus automated market makers

A central limit order book (CLOB) matches traders’ limit orders the way a stock exchange does; the price is the last trade and liquidity is whatever rests near the mid. Kalshi has run a CLOB since launch and Polymarket moved to one in late 2022, keeping only settlement and custody on chain. The alternative, an automated market maker (AMM) such as the logarithmic market scoring rule early Polymarket used, quotes from a formula so there is always a counterparty, but the pool loses to informed traders. AMMs suit thin markets; order books suit 2026 volumes and the professional market makers who want to control their own quotes.

Lifecycle of a market

1. Listing. A CFTC designated contract market (DCM) such as Kalshi self-certifies a contract, which the Commission can stay for review under Dodd-Frank rule 40.11. Global Polymarket’s team writes the question and deploys it on Polygon.
2. Trading. On Kalshi, dollars sit at the clearinghouse. On global Polymarket, positions are tokens backed one-for-one by USDC (see What is a stablecoin?) on Polygon, covered in What is an Ethereum layer 2?.
3. Expiry and resolution. Trading halts, the exchange or oracle determines the outcome, and winning contracts redeem for $1.

Where the money goes

Both venues charge only takers, the side that hits a resting order, with the same formula: fee = rate × contracts × price × (1 minus price). Kalshi’s schedule effective July 7, 2026 sets the standard taker rate at 0.07, with a 0.0175 maker rate switched off by default. Polymarket’s rates vary by category: 0.07 on crypto, 0.05 on sports, 0.04 on finance and politics, zero on geopolitics, per its documentation read in October 2026. Because the formula peaks at 50 cents, a “7%” fee is at most about 1.75 cents per contract. Kalshi reported $263.5m of fee revenue in 2025, 89% from sports, with about 2.9 losing users per profitable one, figures cited by Wikipedia from company reporting: a ratio that looks like a sportsbook, which is the heart of the regulatory argument below.

Resolution: who decides what happened

A contract is only as good as its resolution. Kalshi’s market operations team settles each contract against the source named in its rules. The weakness is interpretation: in March 2026 Kalshi froze about $54m of trades in a market on Iran’s Supreme Leader after Ali Khamenei was killed, citing a policy against paying out on outcomes tied to a death, later conceded the wording was ambiguous and reimbursed about $2.2m, and now faces a proposed class action, per reporting summarised by Wikipedia.

Polymarket and the UMA optimistic oracle

Global Polymarket outsources resolution to UMA’s optimistic oracle. Anyone can propose an outcome by posting a bond, currently $750 in USDC.e per Polymarket’s help pages. The proposal sits in a two-hour challenge window; if nobody disputes it, it stands and the proposer earns a reward. If disputed, the question goes to UMA’s Data Verification Mechanism, a vote by staked UMA holders over a 24-hour commit and 24-hour reveal phase that needs a 65% majority, with the losing bond forfeited and minority voters slashed. It is the token-weighted governance covered in How DAO governance works: honest resolution is meant to be the focal point, but a large enough holder can be the majority.

The March 2025 Ukraine minerals dispute

In late March 2025 a Polymarket market asked whether Ukraine would agree to a minerals deal with the Trump administration before April. No agreement was signed by the deadline, yet after a proposal and dispute the market resolved Yes, with critics pointing to a large UMA holder whose votes were decisive. Polymarket called it a disagreement over the rules rather than a market failure and did not refund traders. A second row came in July 2025 when a market with roughly $240m of volume on whether Volodymyr Zelenskyy would wear a suit saw the oracle outcome reverse mid-dispute. Polymarket’s Protocol V2, announced October 5, 2026, is its first major rebuild of this layer and has published no detail on the oracle. Polymarket US does not use UMA at all: it clears through QC Clearing under exchange rules, so the regulated and global products are best treated as two markets sharing a brand.

Kalshi versus Polymarket: two models, one business

The two leaders reached the same product from opposite directions. Kalshi, founded in 2018, won a DCM licence in November 2020 before launching in July 2021. Polymarket, founded in 2020 by Shayne Coplan, launched on Polygon without a licence, paid a $1.4m CFTC penalty in January 2022 for running an unregistered swap execution facility, blocked US users and grew into the world’s largest venue serving everyone else.

Volume numbers need care. Exchanges quote notional at $1 face per contract, which overstates money at risk: DeFi Rate noted on October 6, 2026 that $84.9bn of combined 30-day notional corresponded to about $22.5bn actually paid, and The Block, whose data Pew used, counts taker notional. Category mix also differs: Pew’s May 2026 analysis found sports were 80% of Kalshi volume from July 2024 to April 2026, crypto 7% and politics 4%, while on Polymarket sports were 39%, politics 32% and crypto 20%. In October and November 2024 politics was 90% of Kalshi volume and 65% of Polymarket’s; it has since become a sports business with an election business attached.

From the 2024 election to the 2026 World Cup

The 2024 US election was the proof of concept. Polymarket took more than $3.3bn on the presidential race by November 5, 2024 and became a fixture of cable coverage; one French trader placed about $30m on Donald Trump across four accounts and collected about $85m. Kalshi, having beaten the CFTC in court in September and October 2024, listed US election contracts legally for the first time.

Pew’s monthly series shows what followed. Combined volume ran between $1.7bn and $2.1bn a month from February to August 2025, jumped to $4.5bn in September 2025 as the NFL season opened, then $8.5bn, $10.1bn and $12.0bn through December 2025. The first five months of 2026 climbed from $17.5bn to $25.7bn, and the FIFA World Cup delivered $47.7bn in June and $53.0bn in July 2026. Kalshi alone reported a record $31bn in June 2026. By August 9, 2026 its 2026 volume had reached $148bn against $23.8bn for all of 2025, per Blockonomi, and the 30 days to October 5, 2026 put it at $66.2bn, up 59% on the prior window as the NFL returned.

Capital followed. Kalshi’s May 2026 round led by TCV valued it at $22bn. Polymarket disclosed $279m of prior funding, including $150m at $1.2bn led by Founders Fund in 2025, before ICE’s investment of up to $2bn on October 7, 2025 at about $8bn pre-money, a deal ICE framed as making it the global distributor of Polymarket’s event data. By October 5, 2026 Polymarket shares on the Forge private market implied roughly $14.3bn, per Prediction News.

Everyone else arrives

Robinhood began routing event contracts to Kalshi in 2025, added NFL and college football in August 2025, and by Q2 2026 prediction markets were its largest transaction revenue line at $156m, ahead of crypto and equities, per Prediction News. Crypto.com, owner of the CFTC-licensed exchange Nadex since 2021, offers sports contracts and is a co-defendant with Kalshi in several state actions. DraftKings launched a predictions product, and Bank of America estimated on October 5, 2026 that it could earn $400m of prediction market fees in 2027. MetaMask integrated Polymarket into its wallet in December 2025, a self-custody channel discussed in Crypto custody explained.

Regulation: the CFTC, the states and the courts

The legal foundation is the Commodity Exchange Act, which gives the CFTC exclusive jurisdiction over swaps on registered exchanges, and Dodd-Frank rule 40.11, which lets it ban event contracts on terrorism, assassination, war, gaming or unlawful activity as contrary to the public interest. For a decade the Commission used that power to keep politics off regulated exchanges, blocking Kalshi’s congressional control contracts in September 2023. Kalshi sued; in September 2024 the District Court for the District of Columbia ruled the CFTC had read “gaming” too broadly, the DC Circuit refused a stay in October 2024 and the Commission dropped its appeal in 2025. The wider map of US agency turf is in SEC vs CFTC crypto regulation.

The sports contract fight

That ruling settled the federal question and opened a state one. When Kalshi listed sports contracts in January 2025, states that licence and tax sportsbooks sent cease-and-desist letters and Kalshi sued them in federal court, arguing CFTC jurisdiction preempts state gambling law. The states have mostly won since. Massachusetts obtained a preliminary injunction with geofencing in January 2026; a federal judge in Ohio ruled on March 9, 2026 that Kalshi’s products are gambling under state law; Nevada’s state court issued a restraining order on March 20, 2026 and Kalshi agreed in July 2026 to halt there; Washington enjoined Kalshi in July 2026; and New York’s attorney general sued the same month seeking around $36bn in penalties.

The CFTC, under chairman Michael Selig, has sued seven states to assert exclusive jurisdiction, and the appellate picture is split. On September 25, 2026 the Sixth Circuit held that Kalshi’s sports event contracts are not swaps under the Commodity Exchange Act, so Ohio and Tennessee may enforce gambling law against them; the Third Circuit, in New Jersey’s case, had taken the opposite approach. Prediction News counted 16 consecutive federal court losses for Kalshi before an Illinois judge granted it a temporary injunction on October 5, 2026. New Jersey, and separately Robinhood and Crypto.com, petitioned the Supreme Court in late September 2026, and on October 6, 2026 Ohio issued cease-and-desist orders to ten platforms.

2026 rulemaking

The Commission has also begun writing rules. It issued advisories on self-certifying event contract series (July 24, 2026) and incentive programmes (August 12, 2026), and a September 22, 2026 staff advisory warned that “mention markets,” contracts on whether a person says a word or attends an event, carry heightened manipulation risk because settlement turns on one person’s conduct. In early October 2026 it sent two event-contract rules to the White House Office of Information and Regulatory Affairs, the step before Federal Register publication; their text is not yet public. Abroad the direction is the reverse: the UK classified prediction markets as betting in March 2026, Spain temporarily banned both platforms in May 2026, MiCA was applied to crypto-based prediction markets from July 2026, and Polymarket is contesting a Dutch ban as of October 2026.

Insider trading and oracle manipulation

Prediction markets reward knowing things first, which is what makes them useful and what makes them dangerous. The first abuse is trading on non-public information. In January 2026 a Polymarket account netted about $400,000 on Nicolás Maduro’s removal from power before any announcement; a US Special Forces soldier was later arrested. The CFTC’s August 28, 2026 order against Gabriel Perez, a White House teleprompter operator, is the first US insider trading enforcement on event contracts: he traded Kalshi mention-market contracts on presidential speeches between December 2025 and February 2026 and was ordered to disgorge $107,539 in profits, pay a $65,000 penalty and stop trading for three years. On September 29, 2026 House Oversight chairman James Comer extended his inquiry from Kalshi and Polymarket to Hyperliquid, Crypto.com and PredictIt’s operator, demanding records on identity checks, employee trades and suspicious activity referrals.

The second abuse is manipulating the outcome or the resolution. Mention markets invite the person being bet on to make the event happen; the Ukraine and Zelenskyy disputes showed a token-weighted oracle can be pushed by a concentrated holder; and markets on wildfires, criticised in July 2026, raise the darker case of a trader with a reason to cause the event. Regulated exchanges answer with surveillance, CFTC Part 16 reporting and a duty to list only contracts not readily susceptible to manipulation; decentralised venues answer with larger bonds and longer windows. Neither has yet shown it can stop a determined insider before the trade.

Are they accurate? What the research says

Berg and colleagues at the Iowa Electronic Markets found that across five US presidential elections from 1988 to 2004 the market beat 74% of contemporaneous polls, and Wolfers and Zitzewitz showed in 2006 that under reasonable assumptions the clearing price equals traders’ mean belief, which is why prices can be read as probabilities at all. Modern data confirms both the strength and the bias: a 2025 microstructure study of more than 300,000 Kalshi contracts found accuracy improved toward expiry but that contracts priced at ten cents or less lost more than 60% on average, so a 10 cent price was closer to a 4% probability. The 2024 election, where both platforms priced a Trump win above the polling consensus and were right, is the showcase, but it is one data point. The practical rule: treat mid-range prices as useful estimates, shade prices below 10 cents down, and treat a thin market with a wide spread as carrying little information.

How institutions and newsrooms use them

Four uses have emerged. Hedging: a fund exposed to a rate decision or a court ruling buys the opposite outcome as insurance; Kalshi said institutional volume rose 800% in the six months before its May 2026 round. Forecasting input: ICE’s stated rationale for its $2bn was to sell Polymarket’s event data to institutional clients as sentiment indicators, and both exchanges now licence feeds to media groups. Editorial: CNN and CNBC show Kalshi odds on air under December 2025 data deals, and Crypto Coin Show’s price predictions page pulls live Kalshi probabilities on 24 markets, including year-end Bitcoin and Ether ranges and when Bitcoin reaches $150,000, with the disclosure that a 27 cent Yes means traders collectively put the odds near 27%; readers can set those against spot data on the CCS crypto prices page. Finally, short-dated price markets act as a cheap alternative to options, though the 15-minute crypto contracts that did $4.8bn on Kalshi in July 2026 look more like the retail perpetuals trade described in Perpetual futures, funding rates and liquidations.

The token question: where POLY stands

Polymarket has signalled since October 2025 that a POLY token is planned, and airdrop speculation has been constant since. As of October 6, 2026 no token has launched, the company has published no eligibility criteria or snapshot date, and the October 5, 2026 Protocol V2 announcement only hinted at it again. Anyone offering POLY, a presale or an airdrop checker today is not Polymarket. Kalshi has no token and none is expected. If POLY arrives, the open questions are whether it carries any claim on fees, whether it plays a role in resolution alongside or instead of UMA, and how users of the regulated US venue could receive it without the exchange becoming a securities issuer.

How we got here: a timeline

Worked example: pricing a $10,000 Bitcoin contract

How to evaluate a prediction market: a checklist

• Who regulates the venue? A CFTC designated contract market with a clearinghouse gives you segregated funds, surveillance and a complaints path; an offshore venue gives you the smart contract and the oracle. Know which you are on, including when one brand has two versions.
• Who resolves the contract, and how? Read the resolution source and dispute path. On Polymarket, check the bond, the two hour window and whether the wording leaves room a token vote could exploit.
• How tight is the spread and how deep is the book? A 2 cent spread with five-figure depth is informative; a 10 cent spread with a few hundred dollars resting is noise. Depth, not headline volume, tells you how far to trust a price.
• What is the all-in cost? Apply the fee formula at your price, add half the spread and compute the breakeven probability as in the worked example. Below 10 cents the favourite-longshot bias matters more than the fee.
• Is the contract exposed to insiders or manipulation? Mention markets and markets on one person’s decisions deserve a discount; the CFTC’s September 22, 2026 advisory says as much.
• Which volume number is being quoted? Notional at $1 face, taker notional and dollars actually paid can differ by a factor of three. Compare like with like.

Risks and open questions

The largest unresolved risk is legal. A Supreme Court ruling that sports event contracts are gambling rather than swaps would strip 80% of Kalshi’s volume, and most of Robinhood’s and DraftKings’ new revenue, of federal protection and hand the question to 50 state regulators. A ruling the other way would make the CFTC the national gambling regulator in all but name, with a statute built for grain futures. The two rules now at the White House may try to draw the line first; either way, the volume figures of 2026 rest on a question no court of last resort has answered.

The second risk is integrity. The 2026 enforcement cases caught people who left obvious trails; the trades that matter are the ones nobody sees. On the decentralised side, oracle design remains unsolved.

The third is the gambling question itself. The National Council on Problem Gambling said in February 2026 that prediction markets carry substantially similar risk to sports betting, and platforms that argue in court that they are financial exchanges market themselves like sportsbooks. Finally, there is concentration: one tracker gave Kalshi 78% of volume in the 30 days to October 5, 2026, so a single clearinghouse, resolution team and fee schedule underpin most of the US market. For a sector built on the wisdom of crowds, it is fair to ask how much of the price is a crowd and how much is a handful of market makers on one venue.

What to watch next

Glossary

Event contract

A binary derivative that pays a fixed amount, usually $1, if a defined event occurs by a deadline and nothing otherwise.

Implied probability

The Yes price expressed as a percentage, before adjusting for fees, spread and bias.

Central limit order book (CLOB)

A matching system where traders post bids and offers and trades occur when prices cross; used by Kalshi and by Polymarket since late 2022.

Designated contract market (DCM)

A CFTC-registered exchange permitted to list futures, options and event contracts for US customers.

Rule 40.11

The CFTC rule allowing it to prohibit event contracts involving gaming, war, terrorism, assassination or unlawful activity if contrary to the public interest.

Optimistic oracle

A resolution system in which a proposed outcome stands unless challenged within a window; UMA’s version secures global Polymarket.

Data Verification Mechanism (DVM)

UMA’s token-holder vote that settles disputed proposals over a 24-hour commit and 24-hour reveal phase, requiring a 65% majority.

Taker fee

The fee charged to the side that executes against a resting order, computed as rate × contracts × price × (1 minus price).

Favourite-longshot bias

The tendency for low-priced contracts to be overpriced and high-priced contracts underpriced relative to true frequencies.

Preemption

The doctrine that federal law overrides conflicting state law; the core of the dispute over state bans on CFTC-regulated sports contracts.

Why it matters

Prediction markets are the first product to come out of crypto and arrive in mainstream finance with its structure intact: a binary token, an order book, collateral held one-for-one and an oracle that decides who was right. What changed between 2024 and 2026 was not the design but the permission: a court ruling made election contracts legal, a licence purchase brought the largest venue back to the US, and a sports season turned a $2bn-a-month curiosity into a $50bn-a-month industry. What remains is whether a sports contract is a swap, whether an exchange can police insiders it cannot see, and whether a token vote can be trusted to say what happened. For anyone reading these prices, the discipline is the same as in any market: know the venue, know the rules, know the cost of crossing the spread, and treat the number as a price set by people with money at risk, which is both its strength and its limit.

Sources

1. CFTC: CFTC Orders Gabriel Perez to Pay $172,000 for Insider Trading of Mention Market Event Contracts, August 28, 2026
2. CFTC: CFTC Releases Staff Advisory on Mention Markets, September 22, 2026
3. Intercontinental Exchange: ICE Announces Strategic Investment in Polymarket, October 7, 2025
4. Kalshi: Fee Schedule, effective July 7, 2026
5. Polymarket: Trading Fees documentation, read October 6, 2026
6. Polymarket: How are prediction markets resolved?, read October 6, 2026
7. UMA: How does UMA’s Oracle work?, read October 6, 2026
8. Pew Research Center: Prediction markets’ trading volume doubled between May and July, largely driven by sports, September 23, 2026
9. Pew Research Center: Kalshi and Polymarket trading volumes dramatically increase since mid-2025, May 27, 2026
10. DeFi Rate: Prediction Market Volume, Kalshi and Polymarket Aggregated Data, October 6, 2026
11. Crypto Briefing: Polymarket gains CFTC approval to launch regulated US prediction markets, November 25, 2025
12. Finance Magnates: Polymarket Buys QCEX Exchange in $112 Million Deal to Reenter the U.S., July 21, 2025
13. Bitcoin Magazine: Polymarket Rolls Out US App Today To Select Users, December 3, 2025
14. The Block: Polymarket discloses past funding rounds before $2 billion ICE investment, October 2025
15. Prediction News: Sixth Circuit rules Kalshi sports contracts not swaps, deepening circuit split, October 5, 2026
16. Prediction News: Supreme Court petitions filed in Kalshi case as Illinois ruling and Missouri order deepen state-federal split, October 1, 2026
17. Prediction News: Polymarket unveils Protocol V2, October 5, 2026
18. Prediction News: House Oversight expands insider-trading probe to Hyperliquid, Crypto.com and PredictIt, September 29, 2026
19. Blockonomi: Kalshi’s 2026 Trading Volume Tops $148B, August 9, 2026
20. Wikipedia: Kalshi, read October 6, 2026
21. Wikipedia: Polymarket, read October 6, 2026
22. Wikipedia: Prediction market, read October 6, 2026

Disclosure: This guide is for education only and is not investment, legal or tax advice.