a marketmaker.cc project

Liquidation Signals

Every forced close on every exchange, the moment it prints. Large orders, cascades, the clusters price is heading into and the whales one move away from liquidation. Over REST and WebSocket.

See a cascade ↓ scroll to play the explainer

A liquidation is a market order nobody chose to send. When thousands of them sit at the same price, they move the market on their own. We see each one as it prints, and we see where the next ones are waiting.

01 / cascade

One print.
Then a thousand.

Scroll. This is how a drop turns into a liquidation cascade, and what we send you while it happens.

  • Longs liquidated as price falls
  • Modelled liquidation levels below the price
  • Shorts liquidated on the bounce
  • Signals sent while it happens

02 / pipeline

From forced order
to signal.

01

Capture

Liquidation streams from Binance, Bybit, OKX, HyperLiquid, Bitget, Gate and other derivatives venues, plus on-chain liquidations from lending protocols. Each print is stamped in µs when it reaches us.

02

Normalize

One record for every venue: coin, side, size in USD at the print price, price, exchange, exchange time and our time. Contract sizes, inverse pairs and symbol names stop being your problem.

03

Aggregate

Rolling totals for 1 minute to 24 hours, by coin, exchange and side. Longs against shorts, the largest single order, how many accounts went down.

04

Map

Open interest, leverage and entry prices give a map of where positions would be liquidated. Large on-chain positions on HyperLiquid and lending protocols are tracked one by one, with their liquidation price.

05

Signal

When a pattern crosses its threshold we send a signal: a large order, a cascade, a cluster cleared, a whale close to its liquidation price, a lopsided market.

03 / signals

Six signals.
Zero screenshots.

Channels post liquidations after the fact. Each signal here is a structured event you can trade, filter or backtest.

Large order

A single liquidation above your size threshold, with venue, side and price. The whale that just got closed.

signal: large_order

Cascade

A burst of same-side liquidations across venues inside a short window. Fires early, updates while the cascade runs.

signal: cascade

Cluster cleared

Price ran through a dense band of modelled liquidation levels. The fuel below is gone, or the next band is close.

signal: cluster_cleared

Whale at risk

A large on-chain position within a set distance of its liquidation price, on HyperLiquid or a lending protocol.

signal: position_at_risk

Imbalance

Longs against shorts over 1h, 4h, 12h and 24h. A 93% long washout reads differently from a balanced day.

signal: side_imbalance

Venue split

Which exchange is doing the liquidating. A move led by one venue is a different trade from a market-wide flush.

signal: venue_share

04 / board

Rolling 24h.
Print by print.

The same data as the API, shown as a board. Sample data replayed from a sell-off day.

total liquidated · rolling 24hsample
$2.41B
LONG $2.24B · 93%SHORT $167.7M
accounts liquidated346,645
largest single order$51.7Mbinance · BTCUSDT
liquidations ≥ $1M

05 / heatmap

Where the next ones
are waiting.

Modelled liquidation levels over six months. Brighter bands hold more positions that would be closed at that price. Move across the map to read a level. Sample model.

BTC · modelled liquidation levels · 6 monthshover the map

06 / access

Every print.
One socket.

Subscribe to raw liquidations, rolling totals or signals only. Filter by coin, exchange, side and size. Reconnect with your last cursor and the gap is replayed, with no duplicates.

History is behind REST with the same fields, back to the first print we stored. Every record carries the exchange time and our receive time in µs.

API preview
// liquidations and signals, filtered on the server
  const ws = new WebSocket("wss://liquidations.marketmaker.cc/api/v1/stream");
  ws.onopen = () => ws.send(JSON.stringify({
    token: "lq_live_•••",
    channels: ["liquidation", "signal"],
    filter: { coins: ["BTC", "ETH"], min_usd: 1000000 },
    resume: "cur_41c7"   // replay what you missed
  }));
  ws.onmessage = e => console.log(JSON.parse(e.data));
wss://…/api/v1/stream · sample events

07 / rules

What is measured,
what is modelled.

Prints are prints

Liquidations come from the exchanges and chains as they happen. The heatmap is a model and is always labelled as one.

Partial feeds flagged

Some venues publish only part of their liquidations, for example one per symbol per second. We mark those records, so totals are never presented as complete when they are not.

Microsecond clock

Exchange time and our receive time are both stored in µs. You can see how late each venue reports.

USD at the print

Size is converted at the liquidation price, so coin-margined and USDT-margined contracts add up.

See the flush before the chart does.

Liquidations, cascades and positions at risk, delivered as data.