# Taker fee

Also written taker, taker fees.

*https://predictionmarkets.tools/glossary/taker-fee · next to Prediction Market Venues*

**Definition:** What a venue charges the side of a trade that removes liquidity, meaning the order that matches immediately against something already resting. The resting side is the maker, and on most venues here the maker pays nothing or is paid. A taker fee is therefore not the cost of trading but the cost of being in a hurry, and it is the only side most schedules publish.

Almost every comparison of prediction market venues is a comparison of taker fees, because the
taker fee is the number venues publish. It is worth knowing what that number is a number about
before you rank anything with it.

## How it works

A trade on an order book has two sides that did different things. One side put an order into the
book and waited — that side supplied the liquidity and is the **maker**. The other side arrived
and took what was sitting there — that side removed the liquidity and is the **taker**. Venues
charge the second and usually not the first, because the first is the one doing them a favour.

The line is drawn on execution, not on order type. Kalshi's fee schedule states that trading fees
are charged only for orders immediately matched with orders sitting on the orderbook, and not for
orders left resting. Limitless puts it in a single sentence — fees apply only to orders that
instantly settle against the book, and a limit order that rests pays nothing. Polymarket's
documentation says makers are never charged. A limit order priced through the spread is a taker
order and will be charged like one, which is the detail that surprises people who believe "limit
order" and "maker" are the same word.

### The maker side is a default, not a guarantee

Kalshi publishes a maker formula next to the taker one, with the same shape and a per-contract
multiplier whose default is zero. That is what a card means by `fees_maker: 0` — nothing is
charged today, under a schedule that already has somewhere to put a charge tomorrow.
[Polymarket US](https://predictionmarkets.tools/tools/polymarket-us) runs the opposite arrangement and pays the resting side, a
rebate applied at the point of the trade, which is why its card carries `fees_maker: -0.31`
rather than a zero. A negative number and a zero are different products, and a listing that
rounded both to "free" would hide the difference.

### A taker fee is not the only way a venue can charge you

[PredictIt](https://predictionmarkets.tools/tools/predictit) has no per-fill charge in either direction. Its money comes from
10% of profit on a profitable sale and 5% of every withdrawal, principal included, so both its
fee fields are `null` and the card's Pricing section carries the rest. A parimutuel pool takes
its cut once, out of the pool, at settlement, so there is no taker to charge — which is why the
`parimutuel` value of `liquidity_model` and a populated `fees_taker` would contradict each other.
And [ForecastEx](https://predictionmarkets.tools/tools/forecastex) charges a flat cent to **both** sides of every execution,
independent of how the contract resolves, which is the one schedule here where the maker/taker
distinction does not exist at all.

## What the catalogue records

`economics.fees_taker` holds the effective fee at fifty cents, not the figure the venue prints.
It has three states and they are three different claims. A number is the rate. `null` means
nothing is charged and renders as "None". **Absent** means a fee exists and we could not price
it — it renders as an em dash, raises a WARN, and is the honest answer where a venue publishes a
percentage without saying what it is a percentage of. Guessing a number into that gap is the one
thing not allowed, because a guessed fee is indistinguishable from a checked one once it is in a
sortable column.

## Why it matters here

**You will compare two published numbers and get it backwards.** Limitless publishes 3.00% and
Kalshi publishes 0.07, and on the same contract at the same price Limitless charges about
fourteen percent less. [What a trade actually costs](https://predictionmarkets.tools/guides/what-a-trade-actually-costs) works
through all four bases in use across this catalogue.

**You will price the entry and forget the exit.** Most of these charges apply again when you
close. A contract held to resolution pays once; a contract traded out of pays twice, plus the
spread in between. Holding to settlement is a fee decision as much as a conviction one.

**You will quote the taker rate for a trade you are not making.** If your orders can rest, the
number that describes you is the maker rate, and on one venue here it is negative. Deciding a
venue is expensive on its taker fee, while intending to make, compares the wrong halves.

**You will treat it as the largest cost.** On a market with real depth it often is. On a thin one
it is not close — the spread and the slippage are, and neither appears in any schedule.
[Where the liquidity comes from](https://predictionmarkets.tools/guides/where-liquidity-comes-from) has the measurements.

## Where you will meet this

- [Adjacent](https://predictionmarkets.tools/tools/adjacent.md)
- [Airavat](https://predictionmarkets.tools/tools/airavat.md)
- [Apify prediction-market scrapers](https://predictionmarkets.tools/tools/apify-prediction-market-scrapers.md)
- [Artemis prediction-market metrics](https://predictionmarkets.tools/tools/artemis-prediction-markets.md)
- [Brier.fyi](https://predictionmarkets.tools/tools/brier-fyi.md)
- [CCXT](https://predictionmarkets.tools/tools/ccxt.md)
- [Confido](https://predictionmarkets.tools/tools/confido.md)
- [Convexly](https://predictionmarkets.tools/tools/convexly.md)
- [Crypto.com Prediction](https://predictionmarkets.tools/tools/crypto-com-prediction-markets.md)
- [DepthFeed](https://predictionmarkets.tools/tools/depthfeed.md)
- [Polymarket dashboards on Dune](https://predictionmarkets.tools/tools/dune-polymarket-dashboards.md)
- [Fatebook](https://predictionmarkets.tools/tools/fatebook.md)

## FAQ

### What makes an order a taker order?

That it matched immediately against something already in the book. Kalshi's fee schedule draws the line exactly there, charging trading fees only for orders that are immediately matched with orders sitting on the orderbook. Limitless says the same in one line, that fees apply only to orders which instantly settle against the book. Nothing about order type decides it — a limit order priced through the spread is a taker order.

### Is the maker side really free?

By default on most venues here, but it is a default rather than a promise. Kalshi publishes a maker formula alongside the taker one, with a per-contract multiplier whose default is zero, which means a named series can carry a non-zero maker fee without the schedule changing shape. Polymarket US goes the other way and pays the maker a rebate at the point of the fill.

### Why does one card say 1.75 and another 1.5 when the venues publish 0.07 and 3.00%?

Because the published figures are in different units and the field is not. economics.fees_taker on every card carries the effective fee at fifty cents, which is the charge on one contract bought at 50 cents as a share of the dollar it pays. Kalshi's 0.07 coefficient works out to 1.75 cents there; Limitless's 3.00% of the fill works out to 1.5 cents on the same contract.

### Does a venue with no taker fee cost nothing to trade?

No. The spread is paid on entry and again on exit, slippage grows with your own order on a curve, and several venues that charge nothing on the fill charge somewhere else instead. PredictIt takes 10% of profit on a profitable sale and 5% of every withdrawal including principal, and has no per-fill charge at all.

## Sources

1. [Fee Schedule for July 2026 - 7.7.26 Update](https://kalshi.com/docs/kalshi-fee-schedule.pdf) — Kalshi, 2026-07-07
2. [Fees](https://docs.polymarket.com/trading/fees) — Polymarket, read 2026-09-19
3. [Fees](https://docs.limitless.exchange/user-guide/fees) — Limitless Exchange, read 2026-09-19
4. [Fee Schedule](https://docs.polymarket.us/fees) — Polymarket US, 2026-09-17
5. [Frequently Asked Questions](https://www.predictit.org/support/faq) — PredictIt, read 2026-09-19

*Last updated 2026-09-21. A reference page, corrected in place — not a dated post.*