A liquidation cascade happens when forced closures trigger more forced closures. Price moves sharply against the side carrying heavy leverage, and those positions are closed automatically. Those closures add pressure in the same direction, so more positions get liquidated.
A move like this is driven by leverage, not always by a change in market structure. So a large candle during a cascade does not automatically mean a new trend.
Step 1: check the Liquidation Feed
The Liquidation Feed card is on Market Ranking and Discovery Intelligence. It shows total liquidation value, the dominant side, the number of events, and the pairs with the largest value.
Watch the direction of the side:
- Sell dominant: long positions are force-closed with sell orders. This adds downward pressure.
- Buy dominant: short positions are force-closed with buy orders. This adds upward pressure.
Check the spread too. Liquidations concentrated in one pair mean something different from liquidations spread across many pairs.
Step 2: check funding
Open Funding Anomaly. This card shows assets whose cost of holding a leveraged position is far from normal. A high score means positions are stacked on one side, and that stacking is the fuel for a cascade.
Step 3: check large trades
Top Trader Flow measures the difference between buying and selling in the largest trades. When the flow runs against the crowded side, that side is more exposed to a squeeze. This card is calculated from trade sizes on the exchange, not from wallet tracking.
Step 4: check the order book and volatility
During a cascade, LBI often reads Liquidity Dry Up or Liquidity Collapse. Volatility Regime also tends to show stressed conditions. Depth Stream shows whether flow is still one-sided or starting to balance.
Reading the end of the move
Large liquidations together with stacked funding can mark the end of a forced move, rather than the start of a new one. But this is a tendency, not a certainty, and the move can run further than expected.
- Wait for liquidations to slow down.
- Wait for the order book to fill back in.
- Read MSI again. A return to a readable regime matters more than the size of the bounce.
Case study: BTC on October 29–30, 2026
BTC fell 5.4% in about seven hours, from 80,810 to 76,444 on Binance Futures. On the first candle of the drop, traded value was about 19 times the previous candle.
| 1-hour candle (UTC) | Low | Close | Traded value |
|---|---|---|---|
| Oct 29 19:00 | 80,735 | 80,810 | US$0.18 billion |
| Oct 29 20:00 | 78,210 | 78,820 | US$3.50 billion |
| Oct 29 23:00 | 77,868 | 77,963 | US$1.58 billion |
| Oct 30 00:00 | 76,717 | 77,208 | US$2.98 billion |
| Oct 30 02:00 | 76,444 | 77,385 | US$1.09 billion |
| Oct 30 06:00 | 78,378 | 78,621 | US$0.42 billion |
| Oct 30 18:00 | 76,270 | 76,615 | US$0.82 billion |
Price and traded value come from BTCUSDT candles on Binance Futures. Card reads come from Antick on the 1D timeframe, for the October 29 candle that closed at 00:00 UTC.
Here is how to read it with the four steps above:
- Liquidation Feed. The card recorded 120.9M across 15,148 events, labeled Sell Dominant. BTCUSDT was only about 20% of the total, followed by SOLUSDT and ETHUSDT, so the pressure was spread out.
- Funding. BTC funding on Binance was 0.01% at 16:00 UTC on October 29, so funding showed no extreme stacking on BTC. Open interest fell 6.4% between 16:00 and 00:00 UTC, from US$7.90 billion to US$7.39 billion.
- Top Trader Flow. Large trades leaned toward selling: Sell 485.16M and Buy 380.06M.
- Order book and volatility. The Volatility card read 12.72 (High), above its 7.79 baseline. Absorption on 1D was labeled Thin.
The first bounce did not mark the end of the move. Price rose 2.85% to 78,621 at 06:00 UTC on October 30, then fell again to 76,270 at 18:00 UTC, below the previous low.
This read supports the suspicion of a liquidation cascade, but does not prove it. Antick does not apply that label, the 1D candle does not cover the drop after 00:00 UTC, and the cards do not show an hourly liquidation history.
Historical context: October 10, 2025
On October 10, 2025, the crypto market saw the largest daily liquidation on record. CoinGlass recorded more than US$19 billion in liquidations, with 85–90% from long positions. The trigger was the US announcement of 100% tariffs on imports from China.
The table below uses 1-hour BTCUSDT candles on Binance Futures. Before the drop, the high on the 13:00 UTC candle was 122,497.
| Candle (UTC) | Low | Close | Traded value |
|---|---|---|---|
| Oct 10 15:00 | 118,424 | 118,965 | US$5.06 billion |
| Oct 10 19:00 | 115,845 | 116,605 | US$2.89 billion |
| Oct 10 20:00 | 112,787 | 114,198 | US$3.63 billion |
| Oct 10 21:00 | 101,517 | 113,254 | US$14.48 billion |
| Oct 10 22:00 | 111,020 | 113,606 | US$3.94 billion |
| Oct 11 12:00 | 111,976 | 112,320 | US$0.40 billion |
Here is how to read it with the four steps above:
- Liquidation side. Long positions being force-closed means sell orders, so this event fits the sell-dominant pattern.
- Stacked positions. CoinGlass noted that long leverage and basis were high before the event. This is the fuel described in step 2.
- Order book. CoinDesk reported that market makers pulled bids and visible depth evaporated. On the 21:00 candle, price fell 17.1% from the peak, and traded value was about 65 times the 10:00–12:00 candle average.
- End of the move. The 21:00 candle closed 11.6% above its low, but that bounce did not start an uptrend. The next day, price still fell to 109,501.
This case also shows two data limits. The low differs between sources: 101,517 on Binance Futures, while the CoinDesk index recorded 106,560. CoinGlass itself estimated the true scale at US$30–40 billion, because exchange liquidation data is incomplete.
Leverage was not the only cause. CoinGlass also noted auto-deleveraging, congestion in transfers between exchanges, and system disruptions at several exchanges. This single case also does not prove that every large liquidation marks the end of a move.
The data above comes from Binance and public reports, not from an Antick read on that date.
Comparison case: BTC 1H, not a cascade
Not every liquidation means a chain reaction. On the candle that closed at 19:00 UTC on October 30, 2026, the Liquidation Feed recorded 15.5M across 1,917 events, labeled Sell Dominant.
This read differs from the October 29 case in three ways:
- Concentrated in one pair. BTCUSDT contributed 10.82M, about 70% of the total.
- A small price move. BTC fell only 0.69% on that candle.
- Calm volatility. The Volatility card read 1.90 (Stable), below its 2.17 baseline.
These numbers come from one hour, so they cannot be compared directly with the 1D figures in the October 29 case. But the pattern is enough to show that a Sell Dominant label alone does not mean a cascade.
Data limits you should know
Antick does not automatically label anything a "liquidation cascade". You conclude that from the combination of cards above.
Liquidation figures are also best read as an indication of direction and scale, not an exact total. Binance's liquidation data stream sends only the latest order per symbol within each 1,000 milliseconds. When liquidations are dense, some events are not delivered.
Related guides: Reading a liquidation cascade, Liquidations, funding anomalies and Top Trader Flow
