Polymarket vs Kalshi vs Limitless

Polymarket is cheapest if you keep funds on the platform. Kalshi is the only fully CFTC-regulated option. Limitless charges nothing on resting limit orders and can charge up to 3% if you take. All three can resolve a market in a way you disagree with, and that is the risk worth reading about first.

Three fee structures that are not comparable

Every venue here prices differently enough that a single "cheapest" answer is misleading. What matters is how you trade.

• Polymarket charges 0% on most markets. Its US-regulated venue applies a flat 0.10% taker fee on the contract premium with 0% for makers, and sports markets added a probability-based taker fee in March 2026 that peaks around 0.75% on a 50/50 contract. There are no deposit or withdrawal fees from the platform itself, though third-party on-ramps charge their own.
• Kalshi charges a variable trading fee based on the contract's implied probability, structured so it is highest at 50 cents and falls toward zero as prices approach 0 or 1. It peaks around $1.75 per 100 contracts. Funding adds a 2% fee on debit card deposits and a flat $2 on withdrawals, with bank transfers free.
• Limitless charges 0.40% flat on AMM trades. On the order book, takers pay 0.40% to 3.00% buying and 0.42% to 1.50% selling, varying with price. Makers pay nothing at all.

The structural insight is that Kalshi and Limitless both charge most at 50/50, which is exactly where the genuinely uncertain and therefore most interesting markets sit. Polymarket's flat approach does not, which is why it wins for active trading on contested outcomes.

The maths on a coin-flip contract

The full comparison

Polymarket

The deepest liquidity in the category, particularly on politics, news and macro events, and 0% trading fees on most markets. It settles in USDC on Polygon, which means funding it is a crypto operation rather than a bank transfer, and there are no platform deposit or withdrawal fees. For an active trader who keeps a balance on the platform, it is comfortably the cheapest.

Two things to know. The March 2026 introduction of probability-based taker fees on sports markets means the 0% headline no longer covers everything, and the US-regulated venue charges 0.10% taker. More importantly, Polymarket resolves through an on-chain oracle with a dispute mechanism, and that mechanism has produced contested outcomes more than once, where holders of a position that most observers considered correct lost on a resolution vote. That is a real, recurring risk and not a theoretical one.

Kalshi

The only fully CFTC-regulated venue here, operating as a designated contract market, and it has opened access to traders in 140-plus countries although it operates without local licences outside the US. Settlement is in dollars over bank rails, so there is no crypto step at all, which for a large group of users is the entire point.

Regulation is not a formality in this category. It means an identifiable supervised entity holds your funds, resolution follows an exchange process rather than a token vote, and there is a regulator to complain to. You pay for that in two places: the variable fee that peaks at 50/50, and funding friction from the 2% debit deposit fee and $2 withdrawals. Bank transfers are free and slower.

Limitless

Crypto-native, on-chain settlement, and a fee structure that rewards exactly one behaviour: resting limit orders pay nothing. If you are willing to post a price and wait, Limitless is free while Polymarket's US venue and Kalshi both charge you. That is a genuine and underappreciated edge for patient traders.

Cross the spread and the picture inverts. Taker fees on the order book run 0.40% to 3.00% buying and 0.42% to 1.50% selling, which at the top of that range is the most expensive execution on this page by a wide margin. Liquidity is thinner than Polymarket's and the market list is more crypto-focused. It is not available to US residents.

Resolution risk is the real risk

Everything above concerns cost, and cost is not what loses people money in prediction markets. Resolution is.

A prediction market pays out based on a written criterion, and the gap between what a market appears to be about and what its resolution text actually says is where losses live. A market titled "Will X happen by June?" may resolve on a specific source publishing a specific figure by a specific timestamp. You can be entirely right about the world and lose because the designated source published a day late, or worded it differently, or the criterion turned on a technicality nobody focused on when the market was created.

The venues differ in how this is adjudicated. Kalshi resolves through a regulated exchange process with defined rules and an identifiable responsible party. Polymarket resolves through an on-chain oracle where disputed outcomes go to a token-holder vote, a design that has produced genuinely contested results where the economically-motivated majority decided the question. Limitless settles on-chain with its own process.

One habit prevents most of this: read the full resolution criteria before entering, every single time, including on markets whose title seems unambiguous. If the criteria are vague, that ambiguity is a position you are taking whether you meant to or not.

What these are actually good for

Two honest use cases and one common mistake.

Hedging defined event risk. Prediction markets price things ordinary markets handle badly: a specific regulatory decision, an election outcome, an approval on a date. Because the payoff is binary and the horizon is defined, sizing a hedge is arithmetic rather than the guesswork an options hedge involves. The constraint is liquidity, which is real on headline markets and thin everywhere else, so a large hedge often cannot be placed without moving the price against you.

Trading a genuine information edge. Headline markets attract attention and are priced efficiently. Niche markets are less efficient, which is where an edge can exist, and also less liquid, which is where exiting becomes the problem. That trade-off is the whole game, and it is the same trade-off as any thin market.

The mistake is treating a probability as a forecast you can beat by reasoning harder. A market at 70% is 70% because informed money put it there, and being confident it should be 80% is the same claim any trader makes about any price. Our explainer on how prediction markets work covers the mechanics, and position sizing applies here exactly as it does anywhere else, with the added wrinkle that a binary payoff means the loss is total rather than partial.

What we could not verify

• Kalshi's exact fee formula. The fee varies continuously with contract probability. The $1.75 per 100 contracts figure is the peak at 50 cents; costs at other price points differ and our worked example is therefore approximate.
• Limitless's price-dependent ranges. Taker fees span 0.40% to 3.00% depending on price, and we used a mid-range estimate. Your actual cost depends on where the contract trades.
• Liquidity by market. Depth varies enormously between a headline election market and an obscure one on the same venue. General statements about liquidity here are not measurements of any specific market.
• Availability. Access rules change frequently and vary by jurisdiction, particularly for US users. Check what is actually available to you before opening an account.

Choosing, in one paragraph

Trade actively, fund with crypto and want the deepest books: Polymarket, accepting oracle resolution risk. Want regulated settlement, dollar rails and recourse: Kalshi, accepting a fee that peaks exactly where the interesting markets are. Patient enough to post limit orders and outside the US: Limitless, where making is free and taking is the most expensive option here. Whichever you pick, read the resolution criteria in full before you enter, and remember the spread usually costs more than every fee on this page combined.