TOTAL VOLUME:
$146.6b
24H VOL:
$161,047,490
24H TRANSACTIONS:
2,781,800,361
OPEN INTEREST:
$1,744,915,607
504,917
Markets across
37,612
events
MATCHED EVENTS:
2,946
PLATFORM COVERAGE:
5
Polymarket:
38%
VS.
Kalshi:
62%
Polymarket isn't fee-free anymore. Here's exactly what it charges, who pays it, and the other three revenue lines behind the business.
Jared Polites
Oct 16, 2026

TL;DR
Traded on Polymarket and never noticed a fee line anywhere on the screen? You're not imagining it, but you're also working from outdated information. Polymarket spent its early years with no trading fee at all, and that reputation stuck even as the platform quietly changed. As of 2026, Polymarket charges an explicit fee on most trades. It's just not charged the way a brokerage or a sportsbook charges one, and that's really the whole story behind how the business makes money.
Yes. Polymarket charges a fee to the taker on most trades, the side of the trade that fills immediately against a resting order rather than waiting in the order book. The rate depends on the market's category, and it doesn't apply evenly across the board.
Geopolitical and world-event markets carry no fee at all. Everything else does, and the rate climbs with how speculative or volatile the category tends to be: political, financial, and tech markets sit at the low end, sports and economic markets in the middle, and crypto markets at the top. If you've seen "Polymarket has no fees" repeated in an older article, including some of PredictionHero's own earlier content, that's the piece that changed.
Polymarket doesn't charge a flat percentage of your trade. It uses a formula: fee equals a category rate multiplied by the number of contracts, multiplied by the price, multiplied by one minus the price. That last part, price times one minus price, is what makes the fee peak at a 50-cent contract and fall off toward $0.01 or $0.99.
The category rates most recently reported are 4% for politics, finance, tech, and mentions; 5% for sports, economics, culture, and weather; and 7% for crypto. Polymarket has revised this schedule more than once in 2026, so treat the exact percentage as current-as-of-writing rather than permanent, and check the fee shown in your own order ticket before trading.
Here's what that looks like on an actual position. Say you put $500 into a crypto market trading at 50 cents, a coin-flip price, buying 1,000 contracts. At the 7% crypto rate, the fee is 0.07 × 1,000 × 0.5 × 0.5, which comes out to $17.50, about 3.5% of the position.
Move the same $500 into a market trading at 90 cents instead, closer to a settled outcome, and the fee drops to roughly $3.50, under 1% of the position. The formula charges the most when the outcome is genuinely uncertain and the least when it isn't.
Nothing. Makers, the side of a trade whose resting limit order gets filled by someone else, pay $0 in fees on Polymarket, full stop. They also get paid for providing that liquidity: Polymarket's Maker Rebates Program returns a share of taker fees, roughly a quarter of what the taker paid, back to the maker whose order filled, paid in USDC.
That asymmetry is deliberate. A market with no one willing to post resting orders doesn't have real prices, just whatever the last person willing to pay the spread was willing to accept. Paying makers to keep orders on the book, instead of charging them, is how Polymarket keeps that book deep enough to trust.
Yes, and this is a split worth knowing about. Polymarket US, the separate CFTC-regulated exchange built from Polymarket's 2025 acquisition of QCX LLC, doesn't use the category-based rates above. It runs one uniform taker rate exchange-wide.
That rate isn't static either. A CFTC rule filing from QCX LLC shows Polymarket US raised its exchange-wide taker coefficient from 5% to 6%, effective July 1, 2026, using the same underlying formula shape (rate times price times one minus price) as the international platform. Makers pay nothing here either. The practical takeaway: a US trader on Polymarket US and an international trader on the main platform can be charged meaningfully different fees on the same event, simply because they're using two different products under two different fee schedules.
Trading fees aren't the only line. Three other mechanisms sit underneath the business, and together they explain why "free to trade" was never quite the same thing as "no revenue."
Yield on deposited collateral. Every open position on Polymarket is backed by stablecoin collateral sitting in a smart contract until the market resolves. At scale, that's a lot of money parked and waiting. In April 2026, Polymarket rolled out its own native collateral token, Polymarket USD (pUSD), backed 1:1 by USDC, replacing the bridged USDC.e it had used before. Reporting on the change points to the same underlying logic banks use on deposits: a native settlement asset gives Polymarket more direct control over the yield that idle collateral generates, instead of that yield accruing somewhere else in the bridging chain.
Institutional data licensing. Polymarket generates a continuous stream of probability data across politics, sports, finance, and culture, built from real money, not a survey sample. In February 2026, Intercontinental Exchange announced it had become the exclusive institutional distributor of that data through a new "Polymarket Signals and Sentiment" product, delivered through ICE's existing market-data infrastructure alongside sources like Reddit and Dow Jones data. That's a real, named commercial relationship, not a speculative revenue line.
A possible future token. Polymarket hasn't launched its own token as of this writing, but reporting through 2026 has pointed to one arriving later in the year, with staking and governance mechanics that could route a share of trading revenue to holders. Treat this as a direction the business appears to be heading in, not a confirmed product.
None of these three require charging more at the point of trade. They're built around what already flows through the platform, capital and data, rather than squeezing the trade itself further. Selling prediction market data is one of the revenue ideas covered here, which is why the quality of that data matters.
Every platform PredictionHero tracks charges takers something; the differences are in the rate, the shape, and what else sits underneath it.
| Platform | Fee model | Makers pay |
|---|---|---|
| Polymarket | Category-based rate × price × (1 − price), 0–7% | $0, plus a rebate |
| Polymarket US | Uniform 6% rate × price × (1 − price) | $0 |
| Kalshi | ~7% rate × price × (1 − price) | 25% of the taker fee |
| Limitless | Flat 0.25% on AMM markets; 0.03–3% on order-book markets | $0 |
| Predict.Fun | Taker-only, collected at settlement; rate not publicly disclosed | $0, plus yield on parked collateral |
| Opinion | 0–1%, rises toward 50% probability, $0.25 minimum | $0 |
Kalshi's structure is the closest direct comparison: both platforms use the same price-times-one-minus-price curve, and both peak around a 50-cent contract. The real difference is on the maker side. Kalshi's makers still pay a reduced fee; Polymarket's pay nothing and get paid instead. Which setup is cheaper for a given trader depends entirely on whether they're usually providing liquidity or taking it, so it's worth checking the live schedule on the specific platform and category before assuming one is uniformly cheaper.
A fee that scales with how uncertain a market is changes behavior at the margins. Trading into a coin-flip market costs real money on the way in; trading a market that's already settled near its outcome barely costs anything. That's a structural nudge toward sizing positions earlier, when the fee is highest and the information edge matters most, rather than chasing a market after most of the uncertainty has already resolved.
It's also a sign of a maturing business, not a bait-and-switch. A platform that only ever made money from nothing eventually can't fund the infrastructure, the liquidity incentives, or the institutional-grade data products that make it worth trading on in the first place. Checking the actual fee on your specific trade, rather than assuming "Polymarket is free" because that used to be true, is the only way to know what a position really costs before you take it. Institutional prediction market data is a different product from the free board you see on the site.
Yes. Takers pay a category-based fee calculated as a rate times price times one minus price, so it peaks at a 50-cent contract and shrinks toward the extremes. Geopolitical markets are fee-free; crypto markets carry the highest rate. Makers pay $0 and earn a rebate.
No, not anymore for takers, though makers still pay nothing. You'll also need USDC to fund a position and a compatible wallet, and Polygon network costs apply to on-chain transfers, separate from the platform's own trading fee.
Beyond taker fees, Polymarket earns yield on the stablecoin collateral backing open positions, licenses its market data to institutions through a 2026 deal with Intercontinental Exchange, and has signaled a future token that could route revenue to holders.
Both use the same rate-times-price-times-one-minus-price formula peaking at 50 cents. The difference is on the maker side: Kalshi's makers pay a reduced fee, while Polymarket's makers pay nothing and receive a rebate instead.
No. Polymarket US runs one uniform rate exchange-wide rather than Polymarket's category-based schedule, currently 6% following a CFTC-filed increase effective July 1, 2026. Makers pay $0 on both products.
PredictionHero aggregates publicly available prediction market data for informational purposes only. This is not financial advice. Prediction markets may not be available in all jurisdictions.
Related

Kalshi, Polymarket, Predict.Fun, and Opinion now all charge takers a fee that peaks on a 50-cent contract and fades toward the extremes.
Jared Polites · Aug 18

Kalshi is CFTC-regulated and dollar-funded. Polymarket runs on USDC with deeper global liquidity and now has its own regulated US arm too. Compare live.
Jared Polites · Jul 30

Yes, Polymarket is a legitimate, well-funded exchange that pays out onchain. Here's the full record, good and bad.
Jared Polites · Sep 25