Prediction Market Venues

What is in this category

The places a contract on a future event is actually listed, matched and paid out. The products listed below disagree on nearly everything that matters — which regulator, if any, stands behind the venue; whether you may open an account at all; whether the winning side is paid in dollars or in a stablecoin; and who gets to say which side won.

They are all bought for the same job. You have a view on something, and you want a position on it. Everything below is what separates one venue from the next once that is settled.

"Real money" is a jurisdiction question, not a product one

This is the single most misread phrase in the category, and it costs readers an afternoon and a KYC submission.

"Real money" reads as a property of the venue — this one pays cash, that one pays points. For the play-money half of the market that is true, and settlement on every card here says so in one word. For everything else it is false. The same venue is a real-money venue in one country and unavailable in the next, and the difference is not something the venue built. It is a line in a regulator's order, a clause a court read differently, or a settlement signed last quarter.

Polymarket is the clearest case. There is one brand and there are two products. The international platform, polymarket.com, is a self-custody order book on Polygon that blocks the United States wholesale; the US platform, polymarket.us, is a CFTC-designated contract market holding dollars at a clearing house, with mandatory identity verification before a first deposit. Same name, different operator, different regulator, different resolution mechanism, different fee schedule. A sentence that says "Polymarket pays real money" is true twice and means two different things.

And the restriction itself is four separate facts that vendors and reviewers routinely merge into one word:

• It blocks by physical location. Polymarket's own geographic-restrictions page, updated 14 August 2026, says permissions follow physical location rather than residency, which is why a restricted-country user who travels can trade and a resident abroad cannot rely on being home.
• It asks where you live at signup. Kalshi's member agreement makes it a representation and warranty — you state that you are not located in one of 55 named jurisdictions, and the agreement can be amended unilaterally on notice.
• It will not move the money. Kalshi serves international users but does not offer them ACH, PayPal or Venmo; Polymarket runs a "close-only" status for Singapore, Poland, Thailand and Taiwan, where existing positions can be closed and no new ones opened, and a separate "hold to resolution" status for Germany, where they cannot be closed at all and the money is locked until the contract settles. The list that actually binds is not the readable one: Polymarket's terms incorporate a separate docs page that carries no date and can change without notice.
• It declines one state. Robinhood's help centre bars Maryland residents from sports event contracts and stops Nevada residents opening new ones, while publishing no list of the states that are fine — the app is the list.

Those are not degrees of the same restriction. A geoblock you can hit before signing up; a withdrawal rail you discover after depositing. Every card in this category says which one it is in its ## Availability section, and dates it, because verified_fields.availability is an error without a date.

"Resolution" is four questions, not one

Resolution is the product. A venue that resolves against its own judgement and a venue that resolves against a named agency release are not the same instrument at any price, and no enum can carry the difference. Four questions, and a card that answers fewer has not described the venue:

1. Who decides. Kalshi's markets team determines the outcome, as the exchange. Polymarket US does the same on its own DCM. Polymarket's international platform does not decide at all — the UMA optimistic oracle does, where anyone may propose an outcome against a bond and a contested proposal escalates to a vote of UMA token holders. Limitless resolves most markets from Pyth Network and its short crypto markets from a Chainlink time-weighted average. Three completely different answers to "who do I complain to".

2. Against which source. Never the title of the market, always the document named in its rules. ForecastEx publishes a terms-and-conditions PDF per contract naming the source agency — for its building-permits contract, the US Census Bureau — and says explicitly that it uses the initial release and ignores later revisions, which is the kind of sentence that decides a position. Kalshi publishes a settlement_sources field per event through its public API, with a name and a URL, which is the most checkable form of this fact anywhere in the category.

3. On what timetable. A market that stopped trading is not a market that has paid. An undisputed UMA proposal settles about two hours after it is made; a disputed one takes days. A venue that resolves by hand generally quotes 24 to 72 hours. ForecastEx pays at a stated clock time on the day of the release. Your capital is locked for the difference.

4. What happens to an ambiguous question. The one nobody advertises, and the one worth reading before it matters. The honest venues publish a fallback: a void-and-refund rule, a dispute window with a bond, an escalation path. Where the fallback is a support ticket, the answer to "what happens when the rules turn out not to cover the result" is "whatever the operator decides", and a refund of your cost basis is not the same as the payout you were owed.

Who is on the other side

The third thing that decides a venue, and the one that is hardest to see from a screenshot, because all four mechanisms render as a percentage.

A central limit order book (CLOB). You trade against another user's resting order; the venue matches and takes a fee. Nobody quotes you a price — the book is the price, and if the book is thin there is no price at all. Polymarket, Kalshi, Limitless, PredictIt and Crypto.com all work this way. The number to look at is depth at the price you want, not the headline volume.

An automated market maker (AMM). You trade against a curve funded by liquidity providers. There is always a price and it always moves against you as you size up, which is a feature for a long-tail market nobody would quote by hand and a cost for a large order. Myriad and Futuur are built this way. Slippage, not the fee, is usually the real cost. Zeitgeist was too, and is the reason this catalogue has an alternatives page for it rather than a card: its Polkadot lease expired in August 2026, so the site and the documentation still answer while the chain beneath them does not.

A parimutuel pool. Nobody quotes anything. Everyone pays into a pool, the pool is divided among the winners after close, and the price you saw while the pool was open was an estimate that keeps moving until it shuts. No real-money venue in this catalogue currently runs one; the shape turns up in scored forecasting products such as Good Judgment Open, where there is no money to divide.

And one that is none of the three. ForecastEx runs a book, but there is no seller. You bid between 1 and 99 cents for the Yes side or the No side, and when a Yes bid and a No bid add up to a dollar the exchange pairs them and becomes the counterparty to both legs. Everything is fully collateralised, so the "other side" is the exchange and the person whose bid completed yours, at once. It is a real mechanism and it does not fit the vocabulary the rest of the category uses.

The economics.liquidity_model field on every card names which of these it is. It is the field to read before the fee, because on an AMM the fee is not where the money goes.

The regulatory split, as of September 2026

Two federal appellate courts have answered the question this category turns on in opposite ways, and neither answer is final. Stated narrowly, from the opinions themselves:

• Third Circuit, 6 April 2026 — KalshiEX, LLC v. Flaherty, No. 25-1922, decided 2–1, precedential. Reviewing a preliminary injunction for abuse of discretion, the court held Kalshi likely to succeed on its claim that the Commodity Exchange Act preempts New Jersey's gaming laws as applied to sports-related event contracts traded on a CFTC-licensed designated contract market, finding those contracts fit the Act's definition of "swaps" and that both field and conflict preemption apply. It affirmed the injunction. This is a likelihood-of-success holding on a preliminary posture, not a final merits judgment.
• Ninth Circuit, 28 August 2026 — KalshiEX, LLC v. Assad, No. 25-7516, decided 3–0, published. Reviewing for abuse of discretion, the court held the contracts are likely not swaps, rejecting the broad reading as not the best textual reading in context, as lacking a limiting principle, and as raising major-questions concerns. It affirmed in part the district court's order dissolving the preliminary injunction against the Nevada Gaming Control Board and Commission, and remanded for the district court to consider Nevada's challenges to Kalshi's election contracts. This is a preliminary-posture holding too, not a final merits judgment. Note the procedural point that gets reported wrongly — the district court dissolved the injunction; the Ninth Circuit affirmed that decision.
• CFTC, 10 June 2026 — the Commission issued a notice of proposed rulemaking, Prediction Markets; Public Interest Determinations, RIN 3038–AF65, published at 91 FR 35806 on 12 June 2026. It would amend Regulation 40.11 and add a new Appendix F to Part 40, specifying which event contracts may be found contrary to the public interest, defining "gaming", and setting out when a contract "involves" one of the activities enumerated in the Act. The comment period closed 27 July 2026.
• Supreme Court — a petition for certiorari from the Third Circuit decision was filed 2 September 2026 and docketed 8 September 2026 as No. 26-299, Flaherty v. KalshiEX, LLC. As of today there is no grant and no conference distribution, and the response is due 8 October 2026.

What this means for a reader is narrower than the headlines suggest. The split is about sports-related contracts and state enforcement against a federally designated exchange. It does not change the status of economic or climate contracts, it does not affect a venue that never served your jurisdiction, and it has not moved a single venue's terms of service by itself — what moves terms is a settlement or a state order, and those arrive without notice. This is exactly why every card here carries a dated availability check rather than a paragraph of analysis.

What to check before you deposit

In this order, because each one can end the exercise:

1. Can you open an account where you physically are, and does the venue enforce that by location, by residency declaration, or only by a clause? The four are different failure modes.
2. What identity check stands between you and a deposit. A connected wallet is not a KYC flow; a regulated exchange is, and it may want proof of address or source of funds on top.
3. How money gets out, not just in. The rail that funded the account is not always the rail that empties it, and some jurisdictions are served for closing positions only.
4. Who resolves, against what, how fast, and what happens if the question is ambiguous.
5. What a fill costs on the mechanism this venue actually runs — the taker coefficient on a book, the slippage on a curve, and whether the maker side is paid or charged.
6. Whether anyone has actually used it. Every card in this catalogue currently says Tested hands-on: no, because funding a dozen venues costs money and we will not claim otherwise. Everything here is read from the venue's own terms, rulebook, fee schedule and documentation, on a stated date.