# Best Perp DEX for Large Orders in 2026

The Perp Reference maintainers · Updated: 24 September 2026 · Readings taken 20 September 2026 · Venue documents read 18 September 2026

## Abstract

The best perp dex for large orders in this window was edgeX: a median $29.5 million rested within
10 basis points of the BTC mid price, and a $1 million order moved it 0.23 bps.
EVEDEX, Lighter and Aster follow. The sequence is the measured depth, not reported volume.

Weights: Order-book liquidity 30 · Market coverage 20 · Trading cost 18 · Security record 14 · Custody and settlement 12 · Access 6. One formula scores every entry, fixed on 22 September 2026 and never adjusted: https://thekiddushclubnft.com/scoring-method

## The best perp dex for large orders, ordered by the depth in the readings

Ordered by the dollar value resting within 10 basis points of the BTC mid price, not by the total; the second table has the depth behind the order.

| No. | Venue | Score of 7 | Maker / taker | Perpetual markets | Model |
|---|---|---|---|---|---|
| 1 | edgeX | 5.3 | 0.040% / 0.045% | 172 | Hybrid |
| 2 | EVEDEX | 4.7 | 0.015% / 0.045% | 52 | Hybrid |
| 3 | Lighter | 5.8 | 0% / 0% | 214 | Hybrid, ZK-proven |
| 4 | Aster | 5.5 | 0% / 0.040% | 578 | Own layer 1, operator-run |
| 5 | Hyperliquid | 5.0 | 0.015% / 0.045% | 324 | On-chain order book |
| 6 | Extended | 5.5 | 0% / 0.025% | 325 | Hybrid |
| 7 | ApeX Omni | 4.7 | 0.020% / 0.050% | 125 | Hybrid |

Facts as of 18 September 2026; fees are the entry tier on the BTC perpetual and market counts are each venue's own. Access differs by country: the terms binding on you are the ones to read before a deposit.

## What the table shows

1. Deepest and dearest are one venue: edgeX held $29.5 million within 10 basis points and charges 0.040% to rest an order, the highest maker fee here.
2. Three of the nine books did not fill all of $1 million: Paradex none of it, Aevo 95%, Hyperliquid 98%.
3. Open interest is not depth: Hyperliquid reported $8.7 billion on 22 September 2026 and held $9.6 million, behind four books here.

## What an order met, book by book

Nine BTC books, read on a ten-minute clock across the two days to 20 September 2026. Aevo and Paradex are measured, not ranked. Margin is the whole of what a perpetual can cost, and the multiple decides how little it takes.

| Venue | Resting within 10 bps | Impact, $100,000 | Impact, $1 million | Share of $1 million filled |
|---|---|---|---|---|
| edgeX | $29.5 million | 0.02 bps | 0.23 bps | 100% |
| EVEDEX | $18.3 million | 2.06 bps | 2.67 bps | 100% |
| Lighter | $16.9 million | 0.60 bps | 1.43 bps | 100% |
| Hyperliquid | $9.6 million | 0.06 bps | 0.60 bps | 98% |
| Aster | $16.1 million | 0.38 bps | 0.92 bps | 100% |
| Extended | $7.4 million | 0.06 bps | 0.78 bps | 100% |
| Aevo | $3.7 million | 3.23 bps | 4.48 bps | 95% |
| ApeX Omni | $1.8 million | 0.38 bps | 3.17 bps | 100% |
| Paradex | $32,811 | 132.47 bps | not filled | none of it |

Read from each public order book, not from reported volume. No account was opened and no order was placed.

## 1. edgeX — the smallest movement at every size

5.3 of 7 · BTC 100x to $1 million · Wallet or MPC login

edgeX matches off-chain and settles on its own EDGE chain, where the documentation describes a
withdrawal made by calling the contracts directly
([edgeX docs](https://pro.edgex.exchange/en-US/home), checked 18 September 2026). It held more resting size than any book here and gave up
0.02 bps, 0.02 bps and 0.23 bps at the
three sizes, least at each. Fees are 0.040% maker / 0.045% taker across 172
contracts.

- In favour: 0.23 bps on $1 million, the least here; A documented withdrawal that skips the operator.
- Noted against: A 0.040% maker fee, the dearest here; 113 readings against 117, the shortest sample here.
- Not for: a resting order paying the maker fee.

## 2. EVEDEX — second on depth, shortest market list

4.7 of 7 · BTC 200x to $50,000 · Wallet signature

EVEDEX matches off the chain; Arbitrum takes position data once enough changes have built up, plus
one write per withdrawal (per the EVEDEX documentation, 18 September 2026). Second-largest resting
size here, 2.06 bps paid on $10,000 and again on
$100,000, 2.67 bps on $1 million. Fees are 0.015% maker / 0.045% taker across
52 contracts, and a $500,000 reserve has stood behind auto-deleveraging since
18 July 2026.

- In favour: $18.3 million within 10 basis points, second here; 2.06 bps on a $100,000 order.
- Noted against: 52 perpetual contracts, fewer than any book here; 2.67 bps on $1 million, wider than five of the six others; No bug bounty is listed for a researcher to claim.
- Not for: a desk whose market is not among the fifty-two.

## 3. Lighter — depth with nothing to pay for crossing

5.8 of 7 · BTC 50x, 2% margin · Wallet signature

Lighter's sequencer matches off-chain and proves every batch on Ethereum, where the deposit
contracts and the state root live. Its book carried the third-largest resting size here and took
1.43 bps out of $1 million against 0.33 bps out of
$10,000. Crossing costs 0% maker / 0% taker on the Standard account, paid for in about
300 milliseconds of taker latency.

- In favour: Nothing to pay on either side of the book; Nine audits, and an Ethereum exit if the sequencer stops.
- Noted against: The free account is slower: 300 milliseconds of taker latency against 140; No public bug bounty, and a 4.5-hour outage on 10 October 2025.
- Not for: a fill needed in the first hundred milliseconds.

## 4. Aster — the tightest quote in the readings

5.5 of 7 · BTC 200x to 400 USDT · Wallet signature or email

Aster matches and settles on Aster Chain, its own proof-of-stake layer 1, live since
16 March 2026. It quoted the tightest spread here, 0.01 bps, a hair behind Lighter
on depth, moving 0.03 bps on $10,000 and 0.92 bps on
$1 million.

- In favour: The tightest spread in the readings, 0.01 bps; 578 perpetual markets, the longest list here.
- Noted against: No external validators in the first phase, and closed chain code; The terms reserve a right to identity verification.
- Not for: anyone who needs the chain checkable from outside.

## 5. Hyperliquid — the largest open interest, a middling book

5.0 of 7 · BTC 40x to $150 million · Wallet signature or email code

Hyperliquid keeps the book on chain: orders, cancels, trades and liquidations happen in HyperCore
on its own layer-1 chain. It reported $8.7 billion of open interest on 22 September 2026,
more than the other six here together, yet its book held only $9.6 million within
10 basis points and moved 0.06 bps on $10,000.

- In favour: 0.60 bps on $1 million, second only to edgeX; A bounty up to 1 million USDC, and the book on chain.
- Noted against: Its listed audits cover the 2023 bridge contract only; Its own vault lost about $4 million in March 2025 and $4.9 million in November 2025.
- Not for: a reader who wants every listed market reviewed.

## 6. Extended — flat fees on a quarter of the depth

5.5 of 7 · BTC 50x to $4 million · Wallet signature, EVM or Starknet

Extended keeps its own book and writes every transaction to Starknet, the collateral in
self-custodial contracts. Its book held a quarter of edgeX's, but impact was flat:
0.06 bps at $10,000 and at $100,000 alike, 0.78 bps at
$1 million. Of 325 active markets, 278 are quoted on request, on an
indicative book.

- In favour: A flat 0% maker / 0.025% taker, with no volume ladder; Up to $500,000 for a critical report, and audited contracts.
- Noted against: 278 markets are quoted on request, with no public book; A market at its open-interest cap turns reduce-only.
- Not for: size in a market with no resting book.

## 7. ApeX Omni — the thinnest book, still filling a million

4.7 of 7 · BTC 100x to $100,000 · Wallet, or email with a Privy wallet

ApeX Omni runs its own book and passes matched trades to zkLink X, whose contracts hold the
deposits and allow a forced exit. It was the thinnest of the seven, $1.8 million within
10 basis points behind a 0.71 bps spread, and moved
0.37 bps on $10,000. It still filled $1 million from resting orders, for
3.17 bps.

- In favour: A forced exit to layer 1 through the zkLink contracts; It filled $1 million from a thin resting book.
- Noted against: $1.8 million within 10 basis points, the thinnest here; 0.050% to take, cut only by volume or staking.
- Not for: an order that must fill in one go at the quote.

## Why depth and fee rank differently

Two venues charge nothing to cross and sit at opposite ends of it. Lighter's Standard account
held a median $16.9 million within 10 basis points; Paradex, also free, held
$32,811, where a $1 million order filled nothing. Aevo, dearest of the
nine at 0.080%, filled 95% of one.

## Quotations

> "Effective February 2, 2026, Maker fees for all VIP tiers were set to 0 bps." — Aster documentation, VIP program, 18 September 2026. https://docs.asterdex.com/program-and-rewards/vip-program.md
> "There is no public order book. The order book displayed on the trading screen is indicative only and does not represent resting orders from other traders." — Extended documentation, RFQ execution, 18 September 2026. https://docs.extended.exchange/extended-resources/trading/rfq-execution
> "Cross margin introduces contagion risk. A large loss in one position reduces available margin across your entire portfolio." — ApeX Omni documentation, Margin and liquidations, 18 September 2026. https://apex-pro.gitbook.io/apex-pro/apex-omni/margin-risk-management-and-liquidations
## FAQ

### How much slippage should I expect on a $1 million perp order?

It depends on the book. A $1 million market order moved the BTC price 0.23 bps
at edgeX, 0.60 bps at Hyperliquid and 3.17 bps at
ApeX Omni, and three of the books read did not fill that size in full from resting orders.

### What is order book depth?

The money sitting in limit orders nobody has taken yet, close to the current price. Here it is
counted in dollars, both sides of the BTC book, out to 10 basis points from the mid — the part a
market order eats first, and a large one runs out of.

### Which perp dex can handle the biggest orders?

Of the nine books we read across the two days to 20 September 2026, edgeX held the most within 10 basis points at a
median $29.5 million, then EVEDEX at $18.3 million and Lighter at
$16.9 million. That is one instrument across two days, not a standing property.

### Does price impact matter more than fees?

At size, usually. A $1 million round trip at the entry tier costs $900.00 at
edgeX and $1,000.00 at ApeX Omni, a hundred dollars apart. Their books
moved 0.23 bps and 3.17 bps on the same order, about
$23 against $317.

### Are zero-fee perp dexes actually cheaper?

Not automatically. Lighter charges nothing on its Standard account and held a median
$16.9 million within 10 basis points, so size crosses cheaply there. Paradex also charges
nothing and held $32,811, where a $1 million order found nothing to fill against.

### Should I split a large crypto order across exchanges?

Splitting helps when one book is the constraint. Our figures show what a single order meets at once:
at $1 million the movement ran from 0.23 bps to 3.17 bps
across the seven ranked here. An order worked in pieces meets a book that refills.

### Does open interest show how liquid an exchange is?

Not reliably. Open interest is reported by the venue; resting depth has to be read off the book.
Hyperliquid reported $8.7 billion on 22 September 2026 against $84.0 million at
Extended, but their books held $9.6 million and $7.4 million within 10
basis points.

### What does a tight spread tell you about a book?

Less than depth does. Aster quoted the tightest spread of the seven here, 0.01 bps, on a
median $16.1 million within 10 basis points, while ApeX Omni quoted
0.71 bps on $1.8 million. A narrow quote says nothing about what
stands behind it.

### How many perpetual markets does a large trader need?

It depends on the book, not on the count. Aster lists 578 perpetual markets and
EVEDEX 52, yet their BTC books were within about three million dollars of each
other. A long list matters when the market you want has resting orders behind it.

### How often do you re-read these order books?

The readings on this entry were taken on a ten-minute clock across the two days to 20 September 2026: a median of
117 readings a book. They are republished when the books are read again, and every figure
carries the day it was taken rather than the day the entry was edited.

### What is price impact on a perp dex?

What a market order pays above the mid price to fill against the book. We measured it by walking
each BTC book at three sizes: a $1 million order moved edgeX 0.23 bps and ApeX
Omni 3.17 bps, while one of the nine books could not fill that size at all.

### How much size can a perp dex absorb before the price moves?

Read the depth, not the reported volume. Within 10 basis points of the BTC mid we found
$29.5 million at edgeX and $1.8 million at ApeX Omni, and $1 million still
filled at both; Paradex, holding $32,811, filled none of it.

## The making of this entry

Nine books, read the same way. Every ten minutes a script asked each venue's public order-book
endpoint for its BTC perpetual and wrote down the dollars resting on both sides inside 10 basis points
of the mid and the cost of crossing it at once at three sizes.
117 readings in one series ending 20 September 2026,
medians throughout. Fees, markets and closed jurisdictions come from each venue's own pages.

One series of readings on one instrument is a sample, not a certificate of liquidity. edgeX answered
113 calls against a median of 117, the shortest sample here.
Price impact assumes the whole order arrives at once. Nothing here is investment advice.

Each venue is linked from its entry heading above. Open interest and the previous day's volume come from CoinMarketCap, captured 22 September 2026. Corrections: corrections@thekiddushclubnft.com.

The Perp Reference maintainers, 24 September 2026

The maintainers are paid to publish this reference, and a venue with an entry here may be the party that paid for it.