Most traders learn what a liquidation is by receiving one, and the description they carry afterwards is usually wrong in a way that matters. It is not the platform closing you at the price you expected. It is a mechanism designed to protect everybody else from your position, and understanding what it is protecting against explains every one of its features that looks unfair.
A leveraged position is money borrowed against collateral you posted. The lender needs the collateral to remain worth more than the loss on the position, at all times, without ever having to ask you for more. When that stops being true, the position is closed by force to stop the shortfall from growing.
This is the whole logic and everything else follows from it. It is not a risk management service offered to you, it is a protection for the party on the other side of your borrowing, which in practice means every other user of the platform. The distinction matters because it explains why the mechanism is indifferent to your view of the trade.
It also explains the timing. A stop loss you place is an instruction that fires at a price you chose, and you can move it or cancel it. A liquidation fires at a price computed from your collateral and the exchange's own risk parameters, and neither you nor anybody at the platform decides in the moment that it should wait.
A liquidation is not a stop the platform places for you. It is a protection against you, and the difference explains everything about how it behaves.
Two margin numbers exist and confusing them is the most common error. Initial margin is what you must post to open the position, and it is the one that determines your leverage. Maintenance margin is the smaller amount that must remain for the position to stay open, and it is the one that triggers a liquidation.
Because the second is smaller than the first, a position can lose a substantial part of its posted collateral without anything happening, and then be closed abruptly when the remaining equity crosses the maintenance threshold. Traders often read the calm of the first phase as safety, which is exactly backwards: the closer the equity gets to maintenance, the less room remains.
The maintenance requirement is also not a fixed percentage across sizes. Platforms raise it in steps as a position grows, on the reasoning that a larger position is harder to unwind without moving the price. A size increase can therefore raise your maintenance requirement and move your liquidation price against you without the market having moved at all.
The estimate shown when you open a position is computed from the state at that moment, and several things move it afterwards without any announcement. Realised funding payments on a perpetual position debit or credit your balance continuously, and every debit brings the liquidation closer. A position held for days can drift meaningfully on funding alone.
Fees do the same. Every fill you take reduces your equity, so an entry built from several fills sits slightly closer to liquidation than the single-fill version of the same position. In isolation each of these is small, which is precisely why they are ignored, and their accumulation over a long hold is not small.
The reference price is the third surprise. Liquidations are usually triggered from an index or mark price built from several sources rather than from the last trade on the venue you are looking at. That is deliberate and it protects you from a single bad print, but it also means the number on your screen is not the number your position is being measured against.
In isolated mode, the collateral attached to a position is only the amount you assigned to it. A liquidation destroys that amount and stops there, and the rest of your balance is untouched. The liquidation price is close, because there is little collateral behind the position, and that closeness is the price of the containment.
In cross mode, the whole available balance backs every open position. The liquidation price is much further away because there is far more collateral behind it, and that distance is genuinely useful. The cost is that a single position failing badly can consume the collateral that was implicitly supporting the others.
Neither mode is safer in the abstract and the choice is a statement about what you are protecting. Isolated protects the account from the position. Cross protects the position from a temporary move, and accepts that the account is the buffer. Traders who lose an account rather than a position have usually chosen cross without framing it that way.
A liquidated position is closed in the market, and the price it actually achieves can be worse than the price at which the liquidation triggered. When that happens the account goes negative, and somebody has to absorb the difference, because the profit on the other side of that trade has already been credited to somebody else.
That is what the insurance fund does. It is a pool built up from liquidations that closed better than their trigger price, and it pays out for the ones that closed worse. In normal conditions it grows, which is why a large fund is often presented as a sign of health rather than what it also is: an accumulation of the difference between trigger and fill.
It is worth knowing whether the platform you use publishes its balance and how it has behaved during past stress. A fund that has been drained and rebuilt has been tested; a fund that has never been touched has not, and its size tells you nothing about what it would do in conditions it has never seen.
If the insurance fund cannot cover the shortfall, the loss has to come from somewhere, and the mechanism used across the market is to close profitable positions on the opposite side of the same instrument. The traders selected are typically those with the highest profit and the highest leverage, which is to say the ones who were most right.
This is the part that surprises people most, and it is not a malfunction. A derivative is a closed system: every gain has a matching loss, and if the losing side cannot pay, the gain cannot be paid either. The alternative would be for the platform to absorb an unbounded liability, which no operator can promise credibly.
Its practical consequence is specific and worth planning for. A position that is working can be closed without your instruction during exactly the kind of violent move that made it work, which means the outcome of a correct trade is not entirely under your control at high leverage. Reducing leverage reduces your rank in the queue as well as your own liquidation risk.
Every gain on a derivative has a matching loss. When the losing side cannot pay, the winning side gets closed. That is arithmetic, not a policy choice.
The obvious cost is the collateral, and it is not the only one. A forced close is executed as an aggressive order that takes whatever liquidity is there, so it pays the wide side of a spread that is usually widest at exactly that moment. That is a real cost and it is invisible in the summary you are shown afterwards.
Most platforms also apply a specific charge on liquidation, distinct from ordinary trading fees, which exists to discourage the practice of using the liquidation engine as a free stop loss. The details vary and the principle does not: closing yourself is cheaper than being closed, always.
The third cost is the one that does the most damage over a career. A liquidation removes the position at the worst available moment and leaves nothing to manage, so a trader who would have reduced, hedged or waited has none of those options. The decision was taken by the mechanism, at the price the mechanism found.
Liquidations are not distributed evenly across prices. They pile up where leverage piled up, which is near the levels that attracted the most positioning, and closing them sends aggressive orders in the direction the price was already moving. Those orders can push the price into the next cluster, which liquidates too.
That is the mechanism behind moves that look disproportionate to any news. Nothing needs to have happened for a cascade to run; it is enough that positioning was concentrated and that the first cluster was reached. Describing this is a description of a mechanism, not a claim that such moves can be anticipated or traded.
What follows from it for a trader is narrow and useful. Placing a position so that its liquidation price sits inside an obvious cluster of other liquidation prices is a choice, and it is available to you to avoid it by lowering leverage. A liquidation price far from where everybody else's sits is far harder to reach by accident.
The first lever is the only one with a large effect, and it is leverage itself. The relationship between leverage and the distance to liquidation is not linear: each step up removes proportionally more of the remaining room than the step before it, and the top of the range gives up almost all of the distance for a comparatively small increase in size.
The second is to place your own exit inside your liquidation price and to treat the gap as the whole point. A stop that fires before the liquidation engine does costs a normal trading fee rather than a liquidation charge, executes as an ordinary order rather than an aggressive unwind, and leaves the remaining collateral where it is.
The third is to recompute rather than assume. Adding to a position, holding through funding, or letting fees accumulate all move the liquidation price, and none of them announce it. The number that mattered when you opened is rarely the number that matters a week later, and reading it again takes seconds.
Three figures describe your exposure and they are all visible. Your current equity, the maintenance requirement of your open positions, and the distance between them. That distance, expressed as a percentage move in the underlying, is the only summary of liquidation risk that means anything, and it is more honest than the leverage multiple.
Comparing it against what the instrument actually does is the step that turns the number into a decision. An instrument whose ordinary daily range routinely exceeds your distance to liquidation is one where the outcome depends on sequence rather than direction, and knowing that before entering is different from discovering it afterwards.
None of this makes a position safe, and it is not intended to. It makes the risk legible, which is a smaller claim and a more useful one. A trader who can state the move that would close them is making a different decision from one who cannot, whatever either of them ends up earning.
No. A stop loss is an instruction you place at a price you chose and can move or cancel. A liquidation is triggered by the platform when your equity falls below the maintenance requirement, at a price computed from your collateral and the platform's risk parameters, and neither you nor the operator decides in the moment that it should wait.
Several things move it silently. Funding payments on a perpetual position debit your balance continuously, fees on every fill reduce your equity, and increasing the size of a position can raise its maintenance requirement in steps. Each is small on its own, which is why they are ignored, and they accumulate.
Initial margin is what you must post to open the position and determines your leverage. Maintenance margin is the smaller amount that must remain for it to stay open, and it is the one that triggers a liquidation. A position can lose much of its collateral quietly and then close abruptly when equity crosses maintenance.
Neither is safer in the abstract. Isolated caps the loss at the collateral assigned to the position but sets the liquidation price close. Cross puts the whole balance behind the position, which moves the liquidation price much further away, and accepts that one position failing badly can consume the collateral supporting the others.
A pool built from liquidations that closed better than their trigger price, used to cover those that closed worse. A liquidation can execute below the price that triggered it, which leaves an account negative, and the profit on the other side has already been credited to somebody. The fund absorbs that difference.
Yes, through auto-deleveraging, when the insurance fund cannot cover a shortfall. The positions selected are typically those with the highest profit and leverage on the opposite side of the same instrument. A derivative is a closed system: if the losing side cannot pay, the winning side cannot be paid either.
Because leverage concentrates at the same levels. Closing one cluster sends aggressive orders in the direction the price was already moving, which can reach the next cluster and liquidate it too. This describes a mechanism; it is not a claim that such moves can be anticipated or traded.
Lower leverage, because the relationship with the distance to liquidation is not linear and the top of the range gives up almost all of the remaining room. After that, place your own exit inside the liquidation price: a stop costs an ordinary fee rather than a liquidation charge and leaves the remaining collateral intact.