Kalshi vs Polymarket: who takes your order and who decides you won

Two instruments that happen to list the same questions

The grid above prints a taker fee of 1.75 against 1.25 and invites the obvious reading: one product, and Polymarket half a point cheaper. Neither half of that holds. Kalshi is a CFTC-designated contract market that checks your identity, holds dollars and decides every outcome itself. Polymarket — the international platform, which is what this page compares — is a self-custody order book on Polygon that checks nobody at signup, holds nothing on your behalf and decides nothing: an oracle does.

What the two share is the mechanism of the fill. Both run a central limit order book where you trade against another user rather than the house, and both price a taker on the same curve shape. Nearly everything else on the two cards points in opposite directions, and the order to read the differences in is the order in which each can end the exercise.

Who will take your order at all

Start here, because for a large share of readers one of the two answers is no.

From the United States, only Kalshi. Polymarket's terms of use, effective 11 August 2026, bar anyone resident, located or incorporated in the US, in capitals, with "THERE ARE NO EXCEPTIONS"; its card reads us_persons: false. Kalshi's reads true, and its member agreement, signup requirements and regulatory notices name no US state at all. The state lists a US reader will have seen — Arizona, Maryland, Massachusetts, Montana, Nevada, New Jersey, Ohio — come from those states' regulators, not from Kalshi, and they are where the question is still live.

Two federal appeals courts have split on it. The Third Circuit, on 6 April 2026, held Kalshi likely to succeed in arguing that the Commodity Exchange Act preempts New Jersey's gaming laws for sports-related contracts on a designated exchange, and affirmed its injunction. The Ninth Circuit, on 28 August 2026, held those contracts likely not swaps and affirmed in part the order dissolving Kalshi's injunction against Nevada's regulators. A petition for certiorari from the Third Circuit decision is docketed as No. 26-299, Flaherty v. KalshiEX, LLC; as the docket read on 4 October 2026 it was pending, with no grant and Kalshi's response due 9 November. Both holdings are preliminary, both concern sports-related contracts and state enforcement, and neither has changed Kalshi's own terms — the category page states them narrowly. None of it touches Polymarket's international platform, which never served the US.

So a US reader is not choosing between these two. The brand's American product is a separate exchange with a different operator — Polymarket US — and the choice among the regulated US doors is a three-way page of its own. Everything below is about the international platform.

Outside the United States, the two lists rarely leave both doors open. At the coarse level of the jurisdictions field Kalshi reads US, EU, Asia and Latin America, and Polymarket reads Asia and Latin America. Underneath, as far as the two cards name countries:

• Closed on both: the United Kingdom, France, Italy, Ireland, Belgium, Australia and New Zealand.
• Open on Kalshi, blocked on Polymarket: Germany, the Netherlands and Japan.
• No new position on either: Singapore, Poland, Thailand and Taiwan are on Kalshi's restricted list and close-only on Polymarket — an existing Polymarket position can be closed, and neither venue opens a new one.
• The reverse case: Kalshi restricts Canada as a whole; Polymarket names only Alberta, British Columbia, Ontario and Quebec.

Both lists are quoted on the cards by example — Kalshi's runs to 55 jurisdictions, Polymarket's to 39 countries plus regions — so for anywhere not named above, read both on the day. Spain and South Korea are off Kalshi's list; Polymarket's card neither names nor clears them.

The two restrictions are different mechanisms. Kalshi's is a representation and warranty you sign at signup — that you are not domiciled, organised or located in a restricted jurisdiction — and not a documented technical block; it restricts trading, not reading, and Kalshi can amend the list unilaterally on notice. Polymarket's follows physical location rather than residency, so a resident abroad is judged by where they are standing, and using a VPN to get round it puts the wallet into close-only mode. Germany shows what that costs in money: there Polymarket can neither open a position nor close one early, and the funds stay locked until the market resolves.

Identity: before the deposit, or only if flagged. Kalshi wants a driver's licence or passport photographed live through its app, with name and date of birth matching exactly, and its help centre says the CFTC may ask for proof of address, employment and source of funds on top. Polymarket asks for a click-through attestation and nothing else in the ordinary flow; an ID and a proof of address in a non-restricted jurisdiction, within 14 days, arrive only when an account is flagged. The second is a default rather than a promise. It can end with the same documents — after your money is already in.

Who decides you were right

Both cards answer this, each in its own terms. What neither can say alone is that the two arrangements trade the same two things in opposite directions: one lets you read the source before you commit, the other lets you contest the result after.

Kalshi decides, as the exchange, and publishes what it decides against. The markets team determines the outcome when a market's resolution criteria are met. Each market's rules name a settlement source, and every event carries it through the public API as a settlement_sources field — a name and a URL. Live examples on 19 September 2026: NATO's own site for the next Secretary General, a NASA GISS data file plus NCEI for a warming market, democrats.org for the next DNC chair; some markets name several news outlets and require them to agree. What you do not get is recourse you can start. There is no bond, no dispute window and no third-party vote; the one second look, the Market Outcome Review Process of Rule 7.1, is opened only by Kalshi, at its sole discretion, and its three-person committee's determination within 24 hours is final. A member can ask for settlement, write to support or report an ambiguity through the bug-bounty form, and if the markets team reads the source differently from you, that is the answer.

Polymarket decides nothing, and the process is open to anyone with a bond. Every market is resolved by the UMA optimistic oracle. An outcome is proposed by posting a bond, typically 750 pUSD — since 5 September 2025 only by whitelisted addresses on markets where Polymarket applies UMA's proposer whitelist. Anyone may dispute within a two-hour challenge period. One dispute sends it to a second proposal round; a second escalates to UMA's Data Verification Mechanism, a vote of UMA token holders, and that vote is final. There is no route from it back to the company, which is a protection and a limit at once: Polymarket cannot overrule a wrong vote any more than it can overrule a right one.

The clock differs too. The category page puts an undisputed UMA proposal at about two hours after it is made and a disputed one at days. Kalshi's contract terms put settlement no later than the day after expiration unless the outcome is under review, its help centre puts determination at one to more than twelve hours after close and "often within about 3 hours", and a review adds at most 24 hours (Rulebook v1.29, Rules 7.1 and 13.1(d), read 9 October 2026). Undisputed, the two are hours apart; disputed, Kalshi's review is a day and UMA's escalation is days.

One more asymmetry, in what the cards record. Kalshi's describes the source as data you can pull; Polymarket's describes the process and not the source — it does not say how a market's resolution source is named or published. Who decides the outcome covers what each arrangement is worth when the rules turn out not to cover the result.

What the fee is a percentage of

Both venues charge a taker on the same formula: a coefficient times contracts times price times one minus price. Per contract, price times one minus price is what you pay times what you stand to win, so the charge peaks at 50 cents, vanishes at both ends and is symmetric about the middle. The coefficient is not a percentage of anything on the screen — 0.07 is not 7% of the trade. That is why the catalogue converts both to one measure, the effective fee at fifty cents: one contract at 50 cents, as a percentage of the $1 a winner is paid. On that basis the cards read Kalshi 1.75, Polymarket 1.25.

Those two numbers are defaults, and they are defaults over different things.

• Kalshi varies the rate by series. The 0.07 coefficient carries a per-series multiplier that defaults to 1, and the schedule lists 155 non-standard series. Some add a maker fee; some — several crypto and political series among them — sit at a multiplier of zero and charge neither side. So 1.75 is the exchange minus a table.
• Polymarket varies the rate by category. 0.07 on crypto, 0.05 on sports, economics, culture, weather and the general bucket, 0.04 on finance, politics, mentions and tech, zero on geopolitics. So 1.25 is one bucket of five.

Put the same question through both and the order can reverse. A crypto contract at a default multiplier costs exactly the same on both, 1.75 at the midpoint, because 0.07 is the coefficient on both — unless its Kalshi series is one of the zero-rated ones, in which case Kalshi is free and Polymarket is not. A politics contract is 1.00 on Polymarket against Kalshi's 1.75, or against nothing if the series is zero-rated. A geopolitics contract is free on Polymarket. The lookup is two documents: the series table on one venue, the category on the other.

The maker fields are not comparable at all, and the two values say why. Kalshi's reads 0: a maker fee exists in the schedule, at a 0.0175 coefficient — 0.44% of the payout at 50 cents — with a multiplier that defaults to zero, and it was switched on for 160 of the 11,078 series the exchange published on 19 September 2026, charged only when a resting order executes. Polymarket's reads None: makers are never charged, and what they receive instead is a rebate printed as a percentage per category — 15% to 25% — paid daily out of a pool the fees fund. That is a share of a pool, not an amount per contract, and no per-contract field can hold it.

Rounding is one more unit trap. Kalshi's 1.75 is $1.75 per 100 contracts. Its schedule documents the round-up twice and not identically — the worked table to the next whole cent, the gloss to a centicent — so a single contract at 50 cents costs 2 cents on one reading and 1.75 on the other. Polymarket's card states no rounding rule. The prediction market fee calculator applies both formulas to your own price and size; what a trade actually costs works through the other units in circulation.

What a win pays out in, and how it leaves

The fee is charged once. The payout asset and the exit rail decide what the win is worth to you in the currency you spend, and here the two cards describe different kinds of money.

Kalshi pays dollars, into an account in your name. Positions are fully collateralised and cleared through Kalshi Klear, only net positions settle, and a simple yes-or-no settlement carries no fee. Getting the money out depends on where you are. In the US, ACH is free both ways; wire withdrawals are documented as not currently supported below $500,000; and Kalshi reserves the right to charge 0% to 2% on any rail. Outside the US the exit is debit card or crypto only — no ACH, PayPal or Venmo — which narrows the dollars-to-a-bank difference for exactly the readers who can use both venues. At year end Kalshi issues tax documents electronically, and trade history downloads with every transaction itemised.

Polymarket pays a token, into a wallet you hold. A winning share redeems for one pUSD, an ERC-20 on Polygon backed one-for-one by USDC and enforced on-chain. Polymarket charges nothing to deposit or withdraw USDC; the only off-ramps its documents name are Coinbase and MoonPay, and the card records what their own pages charge — MoonPay's European disclosure puts a sell by bank transfer at up to 1%, more to a card, with a 3.99 euro minimum — because that step, and its fee, belong to whoever you use for it, not to the venue. There is no account in your name, so nothing arrives at year end; a filer who needs US Form 8949 rows anyway gets them from PolyTaxes, built from the wallet's public history.

How money gets out and who sends you a tax form cover both halves across the category.

The same question, two contracts

Put those sections together and a question listed on both venues is not one contract at two prices. Even where the titles read the same, four things differ:

• The document that decides it. On Kalshi, the source named in that market's rules and published as a field. On Polymarket, the market's rules as read by a whitelisted proposer and, if disputed twice, by UMA token holders. The title settles nothing on either; the rules do.
• The fee bucket. A series on Kalshi, a category on Polymarket — so one politics question can cost 1.75, 1.00 or nothing at the midpoint depending on two separate lookups.
• The asset you are paid in. A dollar cleared at Kalshi Klear, or a pUSD in your own wallet.
• Who you can argue with. Kalshi's markets team, without appeal; or a bonded dispute that ends in a token-holder vote.

Which venue lists the question at all is not something the cards can settle. Kalshi's reports a composition — 200 live open events on 19 September 2026, led by economics at 57 and sports at 55; Polymarket's lists the categories it carries and gives no counts. Neither card gives a volume figure. A claim that one has more of what you want would be invented here; check both.

What this comparison rests on

Both availability blocks were verified on 19 September 2026, and the pricing blocks a day apart — Kalshi's on 20 September, Polymarket's on 19 September — so the two cards are close in date. The documents behind them are not. Kalshi's fee schedule is marked effective 7 July 2026 and had to be read through a reader proxy, because kalshi.com refuses non-browser clients; Polymarket's card gives no effective date for its schedule at all. Kalshi's restricted list dates from an exchange notice effective 22 June 2026. Polymarket's comes from terms effective 11 August 2026 and a help page updated 14 August 2026, and the two disagree with each other — the card says to read the help centre, on the day. The court position is the category page's, as the docket read on 4 October 2026. Availability is the block this catalogue treats as stale soonest, after 90 days, and nobody here has funded an account on either venue.

Which to take, and when to switch

In the United States, take Kalshi; Polymarket's international platform is not an option. The thing to watch is not Kalshi's terms, which name no state, but a state regulator's order or a settlement, which arrive without notice — most of all for sports-related contracts while No. 26-299 is pending.

Outside the United States, let the two lists decide wherever they can. In Germany, the Netherlands or Japan only Kalshi is open; in the UK, France, Italy, Ireland, Belgium, Australia or New Zealand neither is; in Singapore, Poland, Thailand or Taiwan neither will open a new position.

Where both are open, take Kalshi by default. It gives you an account in your name, a settlement source you can read as a field before committing, a year-end tax document and a payout in dollars rather than a token. The two costs to accept are that a resolution cannot be appealed, and that outside the US your money leaves by debit card or crypto rather than by bank transfer.

Switch to Polymarket the moment recourse on the result matters more to you than reading the source first — a bonded dispute inside two hours, ending in a vote the company cannot overturn — or when the specific market sits in a category it charges less for: politics and finance at 1.00 against Kalshi's default 1.75, geopolitics at nothing. Check Kalshi's series table before trusting that second reason, because some of its political and crypto series charge nothing either. What you take on in exchange is a stablecoin in a wallet you hold yourself, no tax form, an identity check that can arrive after the deposit rather than before it, and a conversion to a bank account that is somebody else's fee.