Why we chose Hyperliquid
Not "because it's popular" or "because it has the most volume." HL has three technical properties no other crypto venue has — and those three properties are what makes our non-custodial product even possible.
Three reasons — not marketing
Agent wallets at protocol level
On Binance / Bybit / OKX, to let a bot trade, you issue an API key with trade-only rights. But the key is just a string, the exchange still holds your funds and decides what to do with them. On Hyperliquid an agent wallet is an on-chain entity that literally cannot withdraw — because no code exists for it to do so. That's not exchange policy, that's math.
Builder fee — native economics
Hyperliquid baked builder code into its protocol: every fill routes a share of commission to whoever included the code in the order. AI Traders lives on those 0.05% and takes nothing else from the user. No 20% of profits, no "pay for the algorithm" subscriptions. Other exchanges have no such mechanism — you have to build a custodial model.
On-chain transparency
Every trade and every position on Hyperliquid is publicly visible. That makes copy-trading uniquely possible: a candidate trader's 6-12 months of on-chain history can actually be verified. On Binance — traders show PnL screenshots that prove nothing.
Technical specs
L1 orderbook
Hyperliquid is a custom Layer 1 with a built-in CLOB. Orderbook depth on BTC-PERP rivals top CEX. No chronic slippage like AMM-based DEX.
100-200ms latency
Stated order latency submission-to-confirmation ~200ms. Enough for DCA / Grid. Not enough for HFT — and HFT isn't our scenario.
Funding rate every 8h
Standard perpetual funding every 8 hours. Transparently published. Our models and risk gates use it in real time.
~150 pairs
BTC, ETH, SOL, MAJORS and dozens of altcoins. Diversification room for quant strategies and vault curators.
Mainnet since 2024
Production mainnet has been running over a year, liquidity is stable, protocol has shipped several upgrades without major incidents.
Open API
REST + WebSocket. Full documentation. Official TypeScript SDK, community Python. We use @nktkas/hyperliquid + an in-house Python client.
Comparison with other venues
| Feature | Hyperliquid | Binance | DEX (AMM) |
|---|---|---|---|
| Agent wallet (non-custodial delegate) | |||
| Builder fee (native fee sharing) | |||
| Full on-chain position history | |||
| Perpetuals | |||
| Spot | |||
| Decentralized matching | |||
| KYC required |
What we considered
Before landing on HL we looked at the CEX route (Binance / Bybit / OKX) and DEX alternatives (dYdX, GMX, Drift). Each gives one thing, missing another.
CEX (Binance, Bybit, OKX)
Deep orderbook, low fees, many pairs
Custodial — we or you hold an API key with withdrawal possibility. No builder fee, so the economics turn into "20% of profits," which changes incentives.
dYdX v4
Cosmos-based decentralization, on-chain orderbook
Liquidity 60–70% thinner than HL, higher latency, no agent-wallet equivalent.
GMX, Drift and AMM perps
Full decentralization
AMM order matching = chronic slippage and imperfect price discovery. Not fit for quant strategies with thin edge.
HL today is the only venue combining CEX-grade liquidity, on-chain transparency, and protocol-level non-custodial delegation. Remove any one — our product breaks.
What about the future
If another venue ships a protocol-level agent-wallet equivalent, we'll look. None of the CEX big names are heading there right now (it's technically and regulatorily hard). dYdX v5 hints at something similar — roadmap unclear.
Going multi-exchange would mean rewriting AI Traders as custodial — which means holding API keys, taking those risks, adding in-flight insurance. More expensive, more risk, and the niche is crowded (3Commas, Cryptohopper, Gainium). So the bet for now — deeper into HL, not wider.
Ready to start?
Already on Hyperliquid? Connect to AI Traders in 4 minutes. Not yet? Start with paper mode — just an email needed.
Connect account