Documentation Index
Fetch the complete documentation index at: https://hypernode-docs.polynode.dev/llms.txt
Use this file to discover all available pages before exploring further.
hypernode is the only HyperLiquid API that merges two consensus layers into one WebSocket stream. Every event arrives tagged consensus: "pre" or consensus: "confirmed" so you always know what you’re looking at.
The two layers
| Pre-consensus | Confirmed |
|---|
| Source | Gossip protocol (before validators process) | Confirmed (after block commit) |
| Timing | ~1-2 seconds ahead of public API | ~0.5 seconds behind live |
| What you see | Cancels, modifications, TWAP orders | All orders, fills, cancellations, rejections |
| Completeness | Partial (no order placements, no fills) | 100% of all trading activity |
| Tag | consensus: "pre" | consensus: "confirmed" |
What each layer contains
Pre-consensus events
These are raw actions submitted by traders, intercepted from the gossip protocol before validators commit them. You see what traders are trying to do before it happens.
| Type | What it means |
|---|
cancel | Trader is pulling an order off the book |
cancelByCloid | Cancel by client order ID |
batchModify | Trader is adjusting price or size on multiple orders |
scheduleCancel | Timed cancellation (cancel-after) |
twapOrder | TWAP order submission |
Confirmed events
These are the results after the validator processes the action. Every order placed on HyperLiquid produces exactly one confirmed event.
| Type | What it means |
|---|
order | New resting limit order (status: open) |
order_filled | Order filled immediately on placement |
order_canceled | Order was canceled successfully |
order_rejected | Order rejected (margin, price, balance) |
fill | Trade execution with PnL, fees, direction |
triggered | Stop/trigger order activated (rare) |
Use cases by layer
Market making: pre-consensus cancels = the book is about to change
A burst of pre-consensus cancel events on one side means a market maker is pulling quotes. The order book is about to thin out on that side. You see this 1-2 seconds before the cancels appear in the confirmed stream.
let cancelCount = 0;
let windowStart = Date.now();
ws.on("message", (raw) => {
const event = JSON.parse(raw.toString());
if (event.consensus === "pre" && event.type === "cancel" && event.asset === "BTC") {
cancelCount++;
if (cancelCount > 10 && Date.now() - windowStart < 1000) {
console.log("Cancel storm on BTC — book about to thin");
cancelCount = 0;
windowStart = Date.now();
}
}
});
{
"action": "subscribe",
"filters": {"action_types": ["cancel", "cancelByCloid", "batchModify"], "assets": ["BTC"]}
}
Execution tracking: confirmed fills are your source of truth
Every trade that executes on HyperLiquid produces a confirmed fill event with price, size, direction, PnL, and fees. Use these for accounting, tracking, and analytics.
if (event.consensus === "confirmed" && event.type === "fill") {
const d = event.data;
console.log(`FILL: ${event.asset} ${d.side === "B" ? "BUY" : "SELL"} ${d.sz}@${d.px}`);
console.log(` Direction: ${d.dir} | PnL: ${d.closedPnl} | Fee: ${d.fee} ${d.feeToken}`);
}
{"action": "subscribe", "filters": {"action_types": ["fill"]}}
Whale tracking: confirmed orders show exact sizes
New resting limit orders appear as confirmed order events with the full wallet address, price, and size. Filter by minimum notional to catch whale orders.
if (event.type === "order" && event.consensus === "confirmed") {
const notional = parseFloat(event.data.sz) * parseFloat(event.data.px);
if (notional > 100000) {
console.log(`WHALE: ${event.asset} ${event.data.side === "B" ? "BUY" : "SELL"}`);
console.log(` ${event.data.sz} @ ${event.data.px} ($${notional.toLocaleString()})`);
console.log(` Wallet: ${event.user}`);
}
}
{"action": "subscribe", "filters": {"action_types": ["order"]}}
Liquidation detection: rejected orders signal margin stress
When a trader’s order is rejected with perpMarginRejected or reduceOnlyRejected, they’re running out of margin. A cluster of rejections on one wallet is a liquidation warning.
if (event.type === "order_rejected" && event.consensus === "confirmed") {
const status = event.data.status;
if (status === "perpMarginRejected" || status === "reduceOnlyRejected") {
console.log(`MARGIN STRESS: ${event.user} rejected on ${event.asset} — ${status}`);
}
}
{"action": "subscribe", "filters": {"action_types": ["order"]}}
order_rejected matches when you subscribe to the umbrella type "order". To get only rejections, subscribe to "order" and filter client-side for event.type === "order_rejected".
Copy trading: follow a wallet’s confirmed fills
Subscribe to a wallet address. When you see their confirmed fills, replicate the trade.
const TARGET = "0xd071d6d6ea52f5aa34b79e47f908ee48c8215837";
ws.on("message", (raw) => {
const event = JSON.parse(raw.toString());
if (event.type === "fill" && event.user.toLowerCase() === TARGET.toLowerCase()) {
console.log(`TARGET TRADED: ${event.asset} ${event.data.dir} ${event.data.sz}@${event.data.px}`);
// Execute your copy trade here
}
});
{
"action": "subscribe",
"filters": {"addresses": ["0xd071d6d6ea52f5aa34b79e47f908ee48c8215837"], "action_types": ["fill"]}
}
Combining both layers
The real power is using both layers together. Pre-consensus gives you early warning. Confirmed gives you ground truth.
ws.on("message", (raw) => {
const event = JSON.parse(raw.toString());
if (event.consensus === "pre") {
// Early signal — something is about to happen
handlePreConsensus(event);
}
if (event.consensus === "confirmed") {
// Ground truth — this actually happened
handleConfirmed(event);
}
});
| Strategy | Pre-consensus signal | Confirmed action |
|---|
| Market making | Cancel storm = widen spread | Fill = update inventory |
| Whale tracking | — | Large order = alert |
| Copy trading | — | Target fill = replicate |
| Liquidation sniping | — | Rejection cluster = prepare |
| MEV research | Cancel + modify patterns | Fill = measure impact |
Event volume
At ~6,500 events per second total:
- Pre-consensus: ~1,300/sec (mostly cancels and modifies)
- Confirmed: ~5,200/sec (orders, fills, rejections)
Use filters to narrow the stream. Subscribing to one asset or one wallet drops the volume to a manageable level.