PERPS ATLASPERPETUAL MARKETS INTELLIGENCE
EVIDENCE DESK
Home / Research / Perp Dex Trading Costs Explained
← Research library
TRADING METHODS

How much does a perp DEX trade really cost? Fees, fills and funding

A published maker or taker rate is only one line of a completed perpetual trade. This guide shows how to estimate and then reconcile the costs of entering, holding and closing a position.

By PerpsAtlas Research · Published · Updated · 6 min read

Start with the trade, not the advertised rate

The cost of a perpetual position has three stages: entering, holding and closing. Entry and exit can each involve a trading fee and a difference between the executable price and a chosen reference price. While the position is open, funding can be paid or received. Depending on the venue and account mode, collateral conversion, borrowing, interface fees, withdrawal or liquidation can add further charges. Profit or loss from the market moving is a separate result, not a trading fee.

Notional is the size of the market exposure; margin is the collateral supporting it. If $2,000 margin supports a $10,000 position at 5× leverage, a 0.045% taker rate applies to the $10,000 notional: $4.50 for that fill. Leverage does not reduce the dollar fee. It makes the same fee larger relative to the $2,000 margin. Each later fill needs its own notional and applicable rate because the position size, price or account tier can change.

Sources: GRVT active perp fee schedule and conversion charges ↗

What a $10,000 two-taker round trip says — and what it does not

The official schedules checked on 29 September 2026 list a base 0.045% taker rate for Hyperliquid validator-operated perps, a 0.04% starting taker rate for Aster Crypto General perps, and a 0.045% Level 1 perp taker rate in GRVT's schedule active after 23 March 2026. If both entry and exit are taker fills at an unchanged $10,000 notional, the advertised two-fill amounts are $9, $8 and $9 respectively. The arithmetic is 10,000 × rate × 2. These are examples of stated fee schedules, not measured full trade costs or a claim that a particular account qualifies for them.

Maker means an order adds available liquidity; taker means it trades against liquidity already displayed. A limit order can be a taker if immediately executable, and an order posted as maker may never fill. Hyperliquid's base maker rate is 0.015%; Aster lists 0% maker for Crypto General; GRVT Level 1 lists a small maker rebate. A maker/taker illustration is therefore not a prediction of how a real order will execute.

Contract category matters. Aster's current page separates Crypto General, Crypto Group B, RWA and USD1 perps. It lists a 0.1% starting taker rate for Group B and 1.25 basis points for RWA, effective 7 September 2026. Hyperliquid applies different fee rules to some builder-deployed HIP-3 markets, including deployer settings and growth mode. Aster's Degen mode has distinct charge rules. The $8 Aster example above covers only the documented Crypto General taker rate: whether any additional opening execution charge applies to a particular product is not resolved by that rate page and must be checked in the order preview or account record. Do not reuse the general schedule for a different mode.

Sources: Hyperliquid perp fees and HIP-3 settings ↗ · Aster perp fee categories ↗ · GRVT active perp fee schedule and conversion charges ↗ · Aster Degen mode charges ↗

Define the price benchmark before saying a fill was cheap

A fee advantage of one basis point on $10,000 is only $1. An order can use several price levels, so the best displayed bid or ask is not necessarily the average price for the intended size. To make a reproducible execution comparison, choose the same instrument, order size and side; record near-simultaneous order books; then calculate the size-weighted executable price for each leg. The resulting estimate is still a quote, not a guaranteed fill.

For an actual long round trip, a possible reference is the best-bid/best-ask midpoint immediately before the entry order and a new midpoint immediately before the exit order. Record the two UTC timestamps. Entry execution cost is the paid price above the entry midpoint; exit execution cost is the exit midpoint above the received price, each multiplied by its filled size. Use an explicitly named alternative reference if midpoint is unsuitable for the instrument. This isolates an assumed or measured execution difference from the market price change between entry and exit. Without the reference and timestamps, a statement such as 'two basis points of slippage' is not independently reproducible.

Funding is a holding-cost line, not the whole holding cost

Perpetuals use periodic transfers between longs and shorts. Positive funding commonly means the long pays; negative funding reverses the direction, so funding may be a credit. Hyperliquid documents hourly payments and calculates the payment using position size, oracle price and the interval's rate. Aster describes variable intervals and warns that the actual charge can fall shortly after a stated boundary. GRVT documents market-specific one-, four- and eight-hour schedules. A displayed rate belongs to its interval; multiplying one observation into an annual forecast does not reconstruct a finished trade. Use the settled entries in the account history.

Some collateral modes add borrowing. GRVT's multi-asset margin guide says a negative USDT balance can trigger automatic USDT borrowing with hourly interest until repayment. Converting another asset to repay can add a conversion fee; depositing USDT directly to clear the debt has no such conversion charge under its published rules. These costs are conditional, not a fee charged to every GRVT trader. The ledger should record whether borrowing actually occurred, the hourly charges, the repayment method and the amount converted.

Sources: Hyperliquid funding ↗ · Aster funding rules ↗ · GRVT funding intervals ↗ · GRVT multi-asset margin borrowing ↗ · GRVT active perp fee schedule and conversion charges ↗

A worked ledger with explicit assumptions

Assume a $10,000 Hyperliquid validator-operated perp position supported by $2,000 initial margin. Both fills are taker orders at unchanged $10,000 notional and the documented base rate, making advertised trading fees $9. Assume the buy-entry and sell-exit execution prices, compared with midpoints recorded immediately before their respective orders, create a combined disadvantage of two basis points: $2 on that notional. Assume one funding settlement charges the long 0.01% of a constant $10,000 reference notional: $1. With no builder fee, borrowing, collateral conversion, withdrawal or liquidation, the illustrated cost is $9 + $2 + $1 = $12, before market profit or loss. That is 0.12% of notional or 0.6% of the initial $2,000 margin.

Only the $9 fee line comes from the cited base schedule. The two-basis-point execution difference and 0.01% funding payment are hypothetical inputs, not observations of Hyperliquid or another venue. A real ledger can be higher or lower, and a negative funding payment would reduce cost. One basis point means 0.01%, or $1 on $10,000 notional.

Sources: Hyperliquid perp fees and HIP-3 settings ↗ · Hyperliquid funding ↗

Which extra charges need a separate check?

An interface can add a builder-code fee where a venue permits it. Hyperliquid documents optional builder fees on routed orders. GRVT publishes builder-code terms and chain-specific withdrawal charges. Collateral conversion, borrowing and liquidation have their own rules. Aster's Pro documentation describes liquidation fees in its forced-close process; they do not belong in an ordinary planned-close example. Never add every possible charge to every trade; check the product, account mode, order preview and transaction record first.

For a completed position, keep one row for each fill with timestamp, instrument, side, notional, maker/taker classification, fee rate and charged amount. Keep separate rows for reference midpoint and execution difference, each settled funding transfer, interest, conversions, builder or interface charges, and exit or liquidation charges. Note any rebates as credits. That account-level ledger answers 'what did my trade cost?' Protocol volume, revenue and token buybacks answer different questions and cannot replace it.

Sources: Hyperliquid builder codes ↗ · GRVT active perp fee schedule and conversion charges ↗ · GRVT builder codes ↗ · Aster liquidation rules ↗

What we monitor next

  • Recheck each venue's current contract category, fee tier and extra charges before refreshing the dated schedule examples.
  • Collect synchronized executable quotes before claiming one venue has lower total execution cost.
  • Keep worked assumptions separate from account records and realized funding.

Frequently asked questions

Is a zero maker fee a free perp trade?

No. A maker order may not fill, and a completed position can involve taker fees, execution-price differences, funding, borrowing or other applicable charges.

Does 5× leverage divide my fee by five?

No. The fee applies to the position notional under the venue's rules. Leverage changes the notional-to-margin ratio and therefore the fee relative to collateral.

Can I compare perp DEXs using one posted taker rate?

Only for a narrow fee-schedule illustration with a specified contract and tier. Total realized cost also requires fill prices, settled funding and any account-specific charges.

Compare the evidence

Open the comparison desk →

Related research

Source register

  1. Hyperliquid perp fees and HIP-3 settings · checked 2026-09-29
  2. Hyperliquid funding · checked 2026-09-29
  3. Hyperliquid builder codes · checked 2026-09-29
  4. Aster perp fee categories · checked 2026-09-29
  5. Aster funding rules · checked 2026-09-29
  6. Aster Degen mode charges · checked 2026-09-29
  7. Aster liquidation rules · checked 2026-09-29
  8. GRVT active perp fee schedule and conversion charges · checked 2026-09-29
  9. GRVT funding intervals · checked 2026-09-29
  10. GRVT multi-asset margin borrowing · checked 2026-09-29
  11. GRVT builder codes · checked 2026-09-29
Research revisions
  • 2026-09-29 — First evidence-reviewed edition. The server draft was revised to include conditional GRVT borrowing, an explicit execution benchmark and the limits of Aster's fee-only example.

AI-assisted research checked against cited sources. Facts, assumptions and interpretation are distinguished; this is not a financial audit or a recommendation tailored to you. Editorial standards.

Compare protocols →