# Where an automated strategy fills, and what the fill costs

Updated: 24 September 2026 / Signal Grid research / Venue facts read 18 September 2026 / Order books polled at ten-minute spacing, the last reading dated 22 September 2026

Automation changes which facts matter. A discretionary trader waits for a better price; a strategy
that rebalances every few minutes pays the spread it finds and the fee it was quoted, several hundred
times a month, and nobody is watching when a venue queues withdrawals. So this board is built on
readings rather than features: what the venue has already automated for a trader who writes no code,
what its order interface documents for one who does, what the BTC book held within 10 basis points of
mid when we polled it, what two crossings of the spread cost at the entry tier, what the venue writes
on-chain and when, and what is on public record about the last two years.

The two automation dials were added on 23 September 2026 and are why this board differs from the one
published a day earlier. Between them they hold 38 of the hundred points, exactly what the measured
book and the fee hold together.

The figures come from two places and never from a third. Fees, market counts, settlement design and
access terms are read from each venue's own documentation and public endpoints on 18 September 2026;
what a venue automates and what its order interface documents were read on 23 September 2026.
Depth, spread and price impact are ours, taken at ten-minute spacing, the last reading dated 22 September 2026. Open interest and
24-hour volume are each venue's own numbers, republished by CoinMarketCap on 22 September 2026, which is
why they sit in a column of their own rather than inside a score.

Weights: What the venue automates for you 24 · Programmatic access 14 · Measured liquidity 22 · Round-trip cost 16 · Security and record 12 · Custody and settlement 8 · Market coverage 3 · Access 1. One formula for the whole site, fixed on 23 September 2026 before any venue was marked: https://ellsworthvane.com/rating-method

## What each venue automates for the trader

EVEDEX is the only venue here that runs strategies it wrote itself: four ready-made strategies on the BTC, ETH, SOL and XRP perpetuals, each with a published capacity cap in USDT. On automation it scores 10 of 10; Aster and ApeX Omni, which ship a grid the trader designs, score 6; the vault venues score 3, except Aevo at 4, which publishes a backtest against its own. Paradex documents nothing of the kind and scores nought.

| Venue | Automation, of 10 | What it ships | Score of 40 |
|---|---|---|---|
| EVEDEX | 10 | venue-written strategies, capacity capped | 29.1 |
| Aster | 6 | a grid, rule set by the trader | 30.0 |
| ApeX Omni | 6 | a grid, rule set by the trader | 27.0 |
| Aevo | 4 | a vault to deposit into | 19.5 |
| Lighter | 3 | a vault to deposit into | 28.6 |
| Extended | 3 | a vault to deposit into | 27.5 |
| Hyperliquid | 3 | a vault to deposit into | 24.8 |
| edgeX | 3 | a vault to deposit into | 24.3 |
| dYdX | 3 | a vault to deposit into | 22.3 |
| Paradex | 0 | nothing documented | 20.2 |

The first column is the heaviest dial here and sets these rows; the last is the total of all eight, which does not.

## Venue readings on 18 September 2026

The same panels as one row each, sorted by total. The order is what the weights produce from the
marks, and a venue that placed low lost points somewhere a reader can check.

| # | Venue | Score of 40 | Marks, each out of ten | Maker / taker | Perp markets | Open interest | Volume, 24 hours | Depth, 10 bps | Median spread | Model |
|---|---|---|---|---|---|---|---|---|---|---|
| 1 | Aster | 30.0 | automation 6.0 · api 10.0 · liquidity 6.3 · cost 7.0 · security 10.0 · custody 7.0 · coverage 10.0 · access 8.0 | 0% / 0.040% | 578 | $1.4 billion | $3.2 billion | $14.6 million | 0.01 bps | Own layer 1, operator-run |
| 2 | EVEDEX | 29.1 | automation 10.0 · api 6.0 · liquidity 6.6 · cost 6.5 · security 7.0 · custody 5.0 · coverage 7.8 · access 8.0 | 0.015% / 0.045% | 52 | $704.9 million | $700.5 million | $19.0 million | 4.12 bps | Hybrid |
| 3 | Lighter | 28.6 | automation 3.0 · api 10.0 · liquidity 6.5 · cost 10.0 · security 7.0 · custody 10.0 · coverage 9.2 · access 8.0 | 0% / 0% | 214 | $585.4 million | $2.0 billion | $17.4 million | 0.01 bps | Hybrid, ZK-proven |
| 4 | Extended | 27.5 | automation 3.0 · api 10.0 · liquidity 5.8 · cost 8.5 · security 10.0 · custody 7.0 · coverage 9.6 · access 8.0 | 0% / 0.025% | 325 | $84.0 million | $612.8 million | $7.7 million | 0.12 bps | Hybrid |
| 5 | ApeX Omni | 27.0 | automation 6.0 · api 10.0 · liquidity 4.5 · cost 6.0 · security 7.0 · custody 10.0 · coverage 7.7 · access 10.0 | 0.020% / 0.050% | 125 | $160.2 million | $449.5 million | $1.7 million | 0.68 bps | Hybrid |
| 6 | Hyperliquid | 24.8 | automation 3.0 · api 10.0 · liquidity 5.8 · cost 6.5 · security 7.0 · custody 7.0 · coverage 9.6 · access 8.0 | 0.015% / 0.045% | 324 | $8.7 billion | $13.4 billion | $8.1 million | 0.12 bps | On-chain order book |
| 7 | edgeX | 24.3 | automation 3.0 · api 6.0 · liquidity 6.9 · cost 6.5 · security 7.0 · custody 10.0 · coverage 8.0 · access 8.0 | 0.040% / 0.045% | 172 | $635.5 million | $2.5 billion | $29.3 million | 0.04 bps | Hybrid |
| 8 | dYdX | 22.3 | automation 3.0 · api 10.0 · liquidity 3.9 · cost 6.0 · security 7.0 · custody 7.0 · coverage 5.2 · access 8.0 | 0.010% / 0.050% | 78 | $37.3 million | $87.8 million | $848,659 | 1.97 bps | App-chain, validator-held book |
| 9 | Paradex | 20.2 | automation 0.0 · api 10.0 · liquidity 0.0 · cost 10.0 · security 10.0 · custody 7.0 · coverage 7.0 · access 8.0 | 0% / 0% | 63 | $23.7 million | $21.8 million | $179.00 | 11.21 bps | Hybrid, Starknet appchain |
| 10 | Aevo | 19.5 | automation 4.0 · api 8.0 · liquidity 4.3 · cost 3.0 · security 4.0 · custody 7.0 · coverage 8.5 · access 8.0 | 0.050% / 0.080% | 102 | $12.8 million | $7.7 million | $1.4 million | 2.17 bps | Hybrid |

Marks come from the evidence held for each venue on 18 September 2026. Depth and spread are medians of our own snapshots, taken at ten-minute spacing, the last reading dated 22 September 2026. Open interest and 24-hour volume is a venue self-report, taken from CoinMarketCap on 22 September 2026; market counts and fee tiers come from each venue's own documentation.

## Headline figures

- **Venues on the board:** 10 of 16 — order-book venues we could poll; the rest are listed with their facts and left unscored
- **Round trip on $10,000:** $0.00 to $16.00 — what a taker pays to open and close at the entry tier, 18 September 2026
- **Deepest book on the board:** $29.3 million — median resting depth within 10 bps of mid on edgeX, 117 snapshots
- **Heaviest dial on the panel:** Automation, 24 of 100 — what a venue hands a trader who writes no code; marks ran from nothing to the full ten
- **Open interest behind the board:** $12.3 billion — each venue's own figure as published by CoinMarketCap on 22 September 2026

## How high the BTC multiple runs, and where it stops

What each venue publishes as its highest multiple on the BTC perpetual, and the size at which that
multiple stops. An automated strategy sizes by notional, so the second column matters more than the
first.

| Venue | Cap on BTC | Where the cap holds |
|---|---|---|
| EVEDEX | 200x | $50,000, then 100x to $200,000 |
| Aster | 200x | 400 USDT, then 100x to $800,000 |
| edgeX | 100x | $1 million, then 75x to $3 million |
| ApeX Omni | 100x | $100,000 at 1% margin |
| Extended | 50x | $4 million, then 25x to $8 million |
| Lighter | 50x | 2% margin, no band |
| dYdX | 50x | 2% margin, no band |
| Paradex | 50x | 2% margin, 300 BTC |
| Hyperliquid | 40x | $150 million, 20x above |
| Aevo | 20x | 5% margin, $10 million |

A published cap applies to the smallest position, not to the whole book. Above the notional in the
right-hand column the venue steps the multiple down, which changes the distance to liquidation while a
strategy is still adding. The higher the multiple, the faster a move against the position takes the
margin behind it.

## What one margin balance reaches, and when the book shuts

| Venue | Asset classes | Perp markets | Trading hours |
|---|---|---|---|
| Aster | crypto, stocks, FX, commodities, indices, pre-IPO | 578 | not stated in the documentation we read |
| EVEDEX | crypto, commodities, FX, stocks, indices, pre-IPO | 52 | All markets 24/7, shares and commodities included |
| Lighter | crypto, stocks, FX, commodities, indices, pre-IPO, bonds | 214 | not stated in the documentation we read |
| Extended | crypto, stocks, FX, commodities, indices, pre-IPO | 325 | not stated in the documentation we read |
| ApeX Omni | crypto, stocks, commodities, indices | 125 | not stated in the documentation we read |
| Hyperliquid | crypto, stocks, FX, commodities, indices | 324 | not stated in the documentation we read |
| edgeX | crypto, stocks, commodities, FX | 172 | not stated in the documentation we read |
| dYdX | crypto, commodities | 78 | not stated in the documentation we read |
| Paradex | crypto, stocks, commodities, indices | 63 | not stated in the documentation we read |
| Aevo | crypto, stocks, commodities, FX, pre-IPO | 102 | not stated in the documentation we read |

A venue that lists shares, currencies or metals settles them as perpetual contracts, not as the underlying instrument.

## What a bot meets that a person does not

Two venues on this board quote nothing at all for a round trip, and both charge for it in time.
Lighter's default account trades at 0% maker / 0% taker and adds 300 ms of latency to a taker
order; Paradex charges retail orders from its web interface 0% maker / 0% taker behind a 300 ms
speed bump and bills its programmatic tier instead. A strategy that crosses the spread once an hour
will take that trade happily. A strategy whose edge lives inside the next second cannot.

The second surprise is how little of the cost is the fee. Two crossings of the entry tier run from
$0.00 to $16.00 on $10,000. Price impact on a single $100,000 market
order, measured against the board's books as they stood, ran from 0.02 bps on edgeX to
164.78 bps on Paradex — a gap of several thousand times, while the whole range of
entry-tier fees is sixteen dollars on a $10,000 round trip. That is why measured depth carries 22
points here and the fee carries 16.

## Four things share the name AI trading bot

What a venue calls automation turns out to be one of four products, and the dial worth 24 points
reads the difference. One venue on the board wrote the strategy itself: EVEDEX, a hybrid perpetual
futures exchange that matches off-chain and settles on Arbitrum, runs four of them, on the BTC,
ETH, SOL and XRP perpetuals, all of the same mean-reversion family. The trader chooses a strategy
and a coin, sets the parameters and switches it on; each card carries a trade-frequency label and
a cap in USDT on the capital the strategy will take, and a seven-day backtest sits on the XRP card
(EVEDEX documentation and product page, read 23 September 2026). We publish none of the return figures
printed beside them.

Two venues ship a grid instead, Aster and ApeX Omni, where the tool is the venue's and the rule is
the trader's to design. Six run a vault: capital goes in, an operator or a market-making algorithm
trades it, and there is nothing to configure. Paradex documents none of the three.

Everything else here is a venue you point your own code at, which is why the second new dial asks
what the order interface documents rather than what the marketing page promises: a rate limit
written as a number, and which of post-only, reduce-only and a time-weighted order the venue puts
in writing.

## Every page on this site

**Rankings**

- [AI crypto trading bot platforms, ranked](https://ellsworthvane.com/best-ai-crypto-trading-bot-platforms) — all eight venues ordered by what each one automates, with the total out of 40 beside every row
- [Automated crypto trading platforms](https://ellsworthvane.com/best-automated-crypto-trading-platform) — the same automation order over the seven venues that filled at every size, with the measured impact of three order sizes beside it
- [Exchanges for algorithmic trading](https://ellsworthvane.com/best-crypto-exchange-for-algorithmic-trading) — ordered by measured resting depth within 10 basis points of mid
- [Venues for market-making bots](https://ellsworthvane.com/top-venues-for-market-making-bots) — ordered by what a crossing costs at the entry tier, with the maker side and the measured spread of every book
- [Exchanges with trading bots built in](https://ellsworthvane.com/crypto-exchanges-with-trading-bots) — what each venue automates itself, from a finished strategy to a vault to nothing at all
- [Trading bot fees compared](https://ellsworthvane.com/crypto-trading-bot-fees-compared) — entry-tier maker and taker across the board, and what a round trip costs
- [dYdX alternatives for trading bots](https://ellsworthvane.com/dydx-alternatives-for-trading-bots) — order-book venues for an unattended strategy leaving an app-chain book

**Head to head**

- [EVEDEX vs Lighter](https://ellsworthvane.com/evedex-vs-lighter) — a hybrid book against a zero-fee tier with a latency penalty
- [EVEDEX vs ApeX Omni](https://ellsworthvane.com/evedex-vs-apex-omni) — two hybrids that ship their own automation, priced on depth and on the way out

**Method and background**

- [What is an AI trading bot](https://ellsworthvane.com/what-is-an-ai-trading-bot) — what the label covers, what it does not, and how the claims are usually built
- [How we rate](https://ellsworthvane.com/rating-method) — eight criteria, their weights, the rubric and a worked example
- [About Signal Grid](https://ellsworthvane.com/about) — who writes here, what is read, what is measured and what is paid for

Signal Grid research, 24 September 2026. Corrections: editorial@ellsworthvane.com.

These panels are paid placement.