Crypto Liquidations Explained — Forced Flow, Cascades and Limits
Understand long and short liquidations, forced-flow cascades, open-interest changes and why liquidation data is context rather than an automatic reversal signal.
Read liquidations beside price, positioning and order flow on a live BTC perpetual workspace.
Open live dataWhat liquidations actually measures
A liquidation is forced position reduction initiated by an exchange when collateral no longer satisfies margin requirements. The resulting order is risk control, not a trader choosing that moment to express a new view.
Long liquidations generally create forced selling and short liquidations forced buying. When those orders move price into more vulnerable positions, the process can cascade through successive liquidation levels.
Forced flow can overshoot and later retrace, but it can also accelerate a genuine repricing. The correct question is whether price finds acceptance after leverage is removed, not whether a large liquidation number automatically marks a bottom or top.
Core relationship
Liquidation notional ≈ forced quantity × execution price; displayed aggregates depend on venue coverageThe data contract
Before comparing two values, make sure they answer the same question.
- Some exchanges publish live liquidation events but little or no reliable historical feed.
- A session tape starts when collection starts; it must not be presented as complete daily history.
- Side labels may describe the liquidation order rather than the position that was closed, so adapters must normalize semantics.
- Aggregators differ in venues, deduplication and notional conversion, which explains conflicting totals.
Read it in combinations
| Observed together | Plausible interpretation | What would contradict it |
|---|---|---|
| Price down · long liquidations spike · OI falls | Forced long closure and broad deleveraging amplify the decline. | Fresh OI enters and price keeps accepting lower after the first cascade. |
| Price up · short liquidations spike · OI falls | Short covering provides forced buying but may not create durable demand. | Spot volume builds and price establishes value above the squeeze area. |
| Liquidations large · price barely moves | Available liquidity absorbs forced orders or the aggregate is broad relative to this venue. | Depth disappears and a second wave produces rapid displacement. |
| Cascade ends · CVD remains weak · price reclaims level | Selling aggression no longer creates lower prices, suggesting absorption. | The reclaim fails and leverage rebuilds on the wrong side. |
A reproducible workflow
- 01
Verify feed scope
Record which venue or aggregator is included, whether the feed is live or historical and when collection began.
- 02
Normalize the side
Confirm whether “sell liquidation” means a long position was closed. Venue socket semantics differ.
- 03
Pair with OI change
A genuine deleveraging event should remove exposure. If OI immediately rebuilds, squeeze risk can remain.
- 04
Wait for post-event structure
Use acceptance, reclaim, volume and flow after the cascade. The event is context, not the entry itself.
Common interpretation errors
- Calling every liquidation spike a reversal without checking acceptance afterward.
- Presenting a live session tape as complete historical data.
- Misreading order side as the side of the liquidated position.
- Comparing totals from providers with different venue coverage as if one must be wrong.
Questions
Do long liquidations make price fall?+
They create forced selling that can accelerate a decline, especially when depth is thin. The broader move may have started for other reasons and can continue after the liquidation phase.
Does a liquidation cascade mark the bottom?+
Not reliably. Removing leverage can improve conditions for a reversal, but price must still stop accepting lower and establish a reclaim or other structural evidence.
Why are liquidation totals different across sites?+
Venue coverage, websocket uptime, side normalization, duplicate handling and notional conversion differ. A total without its data contract is not directly comparable.
Related metrics
Apply it in a market playbook
Updated 2026-08-26 · Coverage and refresh behavior are documented in Data sources. Educational research, not financial advice.