Liquidation in Crypto: What It Is and How to Avoid It
Liquidation is the exchange force-closing your leveraged position when your margin falls below the maintenance requirement. It is triggered by mark price, not last traded price, and it costs more than a stop-loss because you pay a liquidation fee on top of the loss. Avoid it by keeping liquidation distance well outside your stop distance.
Liquidation is the one outcome in leveraged trading that is entirely avoidable and entirely unforgiving. It is not bad luck — it is arithmetic that was visible the moment the position opened. This article covers the calculation, the trigger mechanism most traders get wrong, and the escalation ladder that stops it happening.
- Liquidation fires on mark price, an index-derived value — not the last trade on your venue.
- It is strictly worse than a stop: you lose the margin and pay a liquidation fee on top.
- If your liquidation distance is inside your stop distance, your stop is decorative.
- Cross margin risks the whole balance to save one position; isolated margin bounds the loss to that position.
- Adding margin to a losing position postpones liquidation without improving the trade.
What is liquidation, exactly?
When you open a leveraged position you post margin as collateral. As price moves against you, unrealised losses eat that collateral. When the remaining margin falls below the maintenance margin — the minimum the exchange requires to keep the position open — the exchange closes it for you at market.
You are not asked. There is no grace period. And it is materially worse than being stopped out at the same price, for two reasons: liquidation executes at market into whatever liquidity exists at that moment, and most venues charge a liquidation fee on top of the realised loss. On a full liquidation you generally lose the entire margin posted for that position.
How is liquidation price calculated?
The precise formula varies by venue and margin mode, but the structure is consistent. For an isolated long:
liquidation_price ≈ entry_price × (1 − 1/leverage + maintenance_margin_rate)
What that produces in practice, ignoring the maintenance rate for clarity:
| Leverage | Approx. move to liquidation | Room for a 2% stop? |
|---|---|---|
| 2× | ~50% | Ample |
| 5× | ~20% | Ample |
| 10× | ~10% | Comfortable |
| 25× | ~4% | Tight |
| 50× | ~2% | None — stop and liquidation collide |
| 100× | ~1% | Liquidation fires first |
Read the last two rows carefully. At 50× a 2% stop and the liquidation price are effectively the same level, and which one triggers first is a coin flip decided by a wick. At 100× the liquidation is inside the stop, so the stop can never fire — you have a risk control that is mathematically unreachable.
Cross margin changes the picture: the whole account balance backs every position, so liquidation is further away for any single trade but a liquidation event can consume the entire balance rather than one position's margin.
Mark price, not last price — the detail that surprises people
Liquidation triggers on mark price, a value derived from an index of external spot markets plus a funding basis, not the last trade printed on your exchange.
This exists to prevent liquidation cascades being manufactured by pushing one venue's book around. It also means two things traders find counter-intuitive:
- A violent local wick that does not move the broader index may not liquidate you, even though your venue printed a price past your liquidation level.
- Conversely, you can be liquidated at a price your venue never printed, because the index moved.
Always read liquidation distance against mark price. And note the same distinction applies to stop triggers — venues differ on whether stops fire on mark or last price, and sometimes let you choose. A strategy modelled on last-price candles but executed against mark-price triggers is not the strategy you tested.
What actually causes liquidations
Leverage that leaves no room. The dominant cause. At 50× and above, ordinary volatility reaches liquidation before any stop can act.
No stop-loss at all. Either never placed, or placed and rejected — wrong precision, notional under the venue minimum, a transient error — with nothing checking afterwards. This is the failure mode automation is supposed to eliminate and frequently doesn't, because the entry is confirmed and the bracket is not.
Averaging into a loser. Each addition moves the liquidation price closer to spot. A DCA ladder without a hard depth limit converts a bounded loss into an unbounded one. See how insurance orders work and where martingale sizing breaks.
Correlated positions under cross margin. Eight correlated longs sharing one margin pool during a sharp move is a single trade with eight liquidation prices that all approach together.
Funding drain on a quiet position. A crowded position pays funding every interval. Held long enough with thin margin, funding alone can walk you into liquidation with no adverse price move at all. This mode gets almost no attention; the four kinds of liquidation covers it properly.
How to avoid liquidation
1. Size leverage so liquidation is far outside the stop
The rule that subsumes most of the others. If your stop is 2% away, liquidation should be at 8% or beyond — enough that a spike cannot reach it before the stop acts. If you cannot get that separation, the leverage is too high for the stop you need.
2. Place the stop atomically with the entry
Entry and stop are one operation. If the stop is rejected, either the position should not have opened or something must re-place it immediately. Verify the bracket is actually live on the exchange rather than trusting a confirmation message.
3. Prefer isolated margin while learning
Isolated margin bounds the damage to one position's collateral. Cross margin uses the whole balance, which delays liquidation and raises the ceiling on what a single bad trade can cost. Cross is the more capital-efficient tool and the more dangerous one.
4. Cap ladder depth
If you average down, the number of rungs and the total added size must be fixed in advance, and the liquidation price at full ladder depth must be computed before the first entry.
5. Monitor margin ratio, not just price
The number that matters is not distance to liquidation in percent — it is margin ratio, which moves with unrealised P&L, funding and any other positions sharing the pool. Watch that continuously.
A four-level escalation ladder
Rather than a single threshold that either fires or doesn't, graduated response gives you time and choices:
| Level | Trigger | Action |
|---|---|---|
| L1 — Inform | Margin ratio degrading | Notify. No automatic action. |
| L2 — Restrict | Approaching the danger band | Stop adding risk: block new entries and further averaging on that wallet. |
| L3 — Recommend | Materially close | Surface a concrete de-risk action — reduce size, add margin, hedge — for you to accept. |
| L4 — Act | Imminent | Force-close, and only on a wallet explicitly armed for it in advance. |
The design principle worth stealing: the automatic action is opt-in per wallet and off by default. Software that closes real positions without unambiguous prior consent is a worse failure mode than the liquidation it prevents. L1 through L3 are safe to leave on permanently because none of them touch your position.
TradeFloor implements exactly this ladder. Margin health is monitored continuously across all connected wallets; L1–L3 are informational and restrictive; L4 does nothing unless you have armed that specific wallet with an explicit confirmation, and it can be disarmed at any time.
If you are liquidated
Do not immediately re-enter. Post-liquidation is the highest-risk psychological state in trading and the point at which revenge sizing produces a second, larger liquidation.
Instead: write down the liquidation price, your entry, the leverage, and where the stop was — or why there wasn't one. Nearly every liquidation reduces to one of four causes: leverage too high for the stop, stop missing or rejected, ladder deepened without a depth cap, or margin shared across correlated positions. Identify which, fix that specific thing, and reduce size on the next trade regardless of how confident you feel.
Frequently asked questions
What price triggers liquidation — mark price or last price?
Mark price, which is derived from an index of external spot markets plus a funding basis rather than the last trade on your venue. This prevents cascades being manufactured by pushing one exchange's book. It also means a local wick may not liquidate you even if it prints past your level, and you can be liquidated at a price your venue never printed.
At what leverage does liquidation become likely?
Roughly above 25×, and severely above 50×. At 50× liquidation sits about 2% from entry, which collides with a typical 2% stop, so which fires first is decided by a wick. At 100× liquidation is inside most stop distances, meaning the stop can never trigger. Choose leverage so liquidation sits well outside the stop you actually need.
Is liquidation worse than being stopped out?
Yes, in two ways. Liquidation executes at market into whatever liquidity exists at that moment, so the fill is typically worse than a stop at the same level, and most venues charge a liquidation fee on top of the realised loss. On a full liquidation you generally lose the entire margin posted for that position.
Should I use cross margin or isolated margin?
Isolated margin while you are learning, because it bounds the loss to one position's collateral. Cross margin shares the whole account balance, which pushes liquidation further away for any single trade but lets one bad position consume the entire balance. Cross is more capital-efficient and more dangerous.
Can adding margin prevent liquidation?
It postpones it. Adding margin moves the liquidation price further away but does nothing to improve the trade, and it commits more capital to a position that is already wrong. It is a reasonable tactic when the thesis is intact and the position was simply sized too aggressively; it is a trap when used to avoid accepting a loss.
How do I avoid liquidation on a DCA or averaging strategy?
Fix the number of rungs and the total added size in advance, and compute the liquidation price at full ladder depth before the first entry. An averaging ladder without a hard depth cap converts a bounded loss into an unbounded one, because each addition moves liquidation closer to spot.
The Four Kinds of Liquidation, and How to Budget for Each
Margin call, cascade, funding bleed, and the fourth mode nobody talks about — each needs a different defence.
// questions or corrections · [email protected] · more essays · /blog