bboDocs →
Best bid / best ask updates in real time. The lowest-bandwidth way to track top-of-book across perp, spot, and outcome markets.
coinmarketTypes
push fields: coin · time · bbo[best_bid, best_ask] (px · sz · n)
WebSocket · L2 + L4 · Hyperliquid
Hyperliquid orderbook streaming
Real-time L2 and L4 orderbook websockets for Hyperliquid — built for teams demanding low latency and stability. Get an API key and start streaming Hyperliquid orderbooks in under 5 minutes.
~70ms per-block updatesValidator infrastructure
TL;DR
Stream Hyperliquid orderbooks with Hydromancer, either L2 or L4. Less development overhead than running your own node, faster than Hyperliquid native websockets. Our infrastructure is used by the most ambitious market makers and HIP-3 deployers on Hyperliquid.
Streaming infrastructure for the biggest teams on Hyperliquid
BasedKinetiqMasstread.fiVentualsLiquidBullpenSEDADexariSenpiGlobal ComputeAura+65 others
§ L2 vs L4
Two ways to stream — L2 and L4
L2 aggregates resting liquidity at each price, which is the standard orderbook view. L4 exposes the individual orders behind those levels: every resting order with its order ID, side, price, size, and the user address that placed it.
Native Hyperliquid websocket only exposes 20 L2 levels and sends updates every 500ms. Hydromancer's l2Book channel reconstructs the full book on the client through snapshot + delta sync, with per-block deltas every ~70ms.
HYPE / USDL4~70ms
pricesizeuser
21.268132.940x7f3...a9c2
21.267143.500x3d1...f8b4
21.266537.100xa21...c7e9
spread 0.001 (0.005%)mid 21.2535
21.253130.120x8e4...d2a1
21.2490.580xf12...b8c3
21.248137.630x5c9...e1f7
§ WebSocket channels
Three streams, the whole orderbook
Subscribe by sending {"method": "subscribe", "subscription": {"type": "<channel>", "coins": [...]}}. Each channel pushes per-block, with a per-subscription seq for ordering and recovery.
l2BookDocs →
Full L2 snapshots at configurable depth, sent every block (~70ms). The standard book view — price-level aggregation, no individual orders.
coinsnSigFigsmantissanLevelsmarketTypes
push fields: levels (bids/asks · px · sz · n), time, seq
l4BookUpdatesDocs →
Order-level updates with the user address on every resting order. Combine with our L4PerpBookSnapshot REST call for full local-book sync. Hydromancer-only.
coinsmarketTypes
diffs: type (new · update · remove) · oid · user · side · px · sz
Snapshot + delta sync is handled by the Hydromancer Python SDK (L4OrderbookClient). Coverage includes Hyperliquid crypto perps, HIP-3 perps, and HIP-4 markets.
§ Alternatives
Three ways to stream the Hyperliquid orderbook
Every team building on Hyperliquid faces the same choice early on. Each path trades off latency, depth, ops burden, and engineering time. Here's the honest breakdown.
Option 01Free
Native Hyperliquid websockets
Subscribe to the official Hyperliquid websocket directly. The fastest way to ship a prototype.
Good for
- — Lightweight bots and prototypes
- — Small projects, low subscription count
- — Learning the schema
Breaks down at
- L4 visibility (not exposed)
- Capped orderbook depth
- Best-effort only, no guaranteed SLAs
Option 02DIY
Run your own Hyperliquid node
Peer with the network, ingest orderbook events at the source, and own your pipeline end-to-end.
Good for
- — Maximum control of the pipeline
- — Heavy internal customisation
- — Teams with platform engineers to spare
Breaks down at
- Peer discovery and node maintenance
- Replay and state recovery
- On-call burden during network instability
- Weeks of engineering
Option 03Recommended
Stream with Hydromancer
Validator-peered infrastructure, L4 with user addresses, snapshot + delta sync handled by our SDK, and high-fanout subscriptions out of the box.
You get
- — —
l2Book,bbo,l4BookUpdates - — — L4 with the
useraddress on every order - — — Multi-market fanout without falling off the socket
- — — HIP-3 deployers covered with the same schemas
Trade-off
- — You're using a vendor
Most teams start on option 1, burn a few weeks on option 2, and end up on option 3. We built Hydromancer because we wanted to skip the first two steps.
§ Head-to-head
Native websockets vs. own node vs. Hydromancer
The honest side-by-side. Each option has a reason to exist; here's where each one lands across the things that actually matter at production scale.
Concern
Native websocket
Own node
Hydromancer
Order-level visibility
Seeing the maker behind each resting order.
✕L2 only — 20 aggregated price levels.
✓L4 book updates with
user on every order.✓
l4BookUpdates with user on every order.Orderbook depth
How deep into the book you can subscribe.
✕20 L2 levels max.
✓Full depth from the node.
✓Configurable depth (
nLevels) plus full L4.Snapshot + delta sync
Maintaining a correct local book under reconnects.
✕You write the buffering and replay logic.
✕You also write it — plus the ingestion pipeline.
✓Handled by the Hydromancer SDK (
L4OrderbookClient).Multi-market scale
Subscribing to many markets at once.
✕Rate limited.
~Scales with your infra investment.
✓Built to stream many markets at once across Hyperliquid.
Engineering time to first byte
How long until live data hits your code.
✓Minutes.
✕Days to weeks.
✓Under 5 minutes — API key + subscribe.
Ongoing ops burden
Who's on-call when the network gets weird.
~None, but without support either.
✕You are on-call.
✓We are on-call. Direct line to the founders if anything's off.
§ Built for
Built for the teams pushing Hyperliquid forward
L4 visibility, low latency, and stability are the three things every serious Hyperliquid product eventually needs. We ship all three by default.
01
Trading terminals
Render the book your users actually want — bids, asks, addresses, queue position — without writing your own sync.
02
Market makers
Track queue position, watch competitor flow, and react to BBO changes per block.
03
HIP-3 deployers
Attribute LP rewards and monitor maker activity using the per-order user field on every resting order.
04
Copytrading platforms
Watch thousands of wallets simultaneously without falling off the websocket every five minutes.
05
Analytics & explorers
Real-time depth, spread, and orderflow surfaces for dashboards and public explorers.
06
Quant research
Pair live streams with our Reservoir historical archive for matched train-test pipelines.
§ One stack
Streaming that composes with the rest of the stack
80+ apps and trading firms use Hydromancer for their Hyperliquid data ops. One API key covers real-time L2/L4 streams, on-demand REST, HIP-4 prediction markets, and the HIP-3 stack.
For general websockets and Hyperliquid data streaming, read the Hyperliquid data streaming page.
§ FAQ
Frequently asked questions
What is Hyperliquid orderbook streaming?+
Hyperliquid orderbook streaming allows you to get all bids/asks and book differences of any Hyperliquid, HIP-3, or HIP-4 market. We offer it at a lower latency than anybody else.
What is the difference between L2 and L4 orderbooks on Hyperliquid?+
L2 shows you the aggregated bids and asks at every level. L4 also shows you an address per each bid/ask, which is a special feature of onchain DEXes.
How is Hydromancer's orderbook stream faster than the native Hyperliquid websocket?+
We are running our own validator node and also have performed extensive software optimizations.
Why can't I just use the native Hyperliquid websocket API?+
You can use native Hyperliquid websocket — we provide a lower latency, more stable, improved alternative to the native WS.
Do I need to run my own Hyperliquid node to get full orderbook depth?+
Yes, to get full orderbook depth teams usually run their own node. Or just use Hydromancer.
Does Hydromancer cover orderbooks for HIP-3 deployers and HIP-4 markets?+
We cover all HIP-3 deployer orderbooks, including Trade[XYZ], and all HIP-4 markets.