The number that gets people is not the leverage, it is the distance to liquidation, and those two are not the same thing. Someone opens a 20x long and assumes they have a 5 percent cushion before things go wrong, because 100 divided by 20 is 5. Then price drops 4 percent and the position is already gone. The gap between what the leverage implies and where liquidation actually sits is where most of the surprise lives, and almost all of it is knowable before you click buy. So let me walk through how I compute the liquidation price, why it drifts closer than you think on larger positions, and what actually happens when you get there.
The base formula, isolated margin
Start with the simplest case. You are on isolated margin, meaning only the margin you assigned to this one position can be used to keep it alive. Nothing else in your account bails it out. For a long, the liquidation price is roughly the entry price minus the entry price times your initial margin fraction, plus a small correction for the maintenance margin. In plainer terms, liquidation happens when your losses eat through your margin down to the point where only the maintenance requirement is left.
A cleaner way to hold it in your head. Take your entry price. Your initial margin fraction is one divided by leverage, so 20x gives you 0.05. Your maintenance margin fraction is some smaller number the exchange sets, often something like 0.005 at the base tier. For a long, liquidation sits near entry times one minus initial margin fraction plus maintenance margin fraction. For a 20x long with a half percent maintenance requirement, that is entry times one minus 0.05 plus 0.005, so entry times 0.955. Price only has to fall about 4.5 percent, not 5, before you are liquidated. The maintenance margin ate half a percent of your cushion before the trade even started.
For a short, flip the sign. Liquidation is above your entry, at roughly entry times one plus initial margin fraction minus maintenance margin fraction. The asymmetry is worth internalizing, because a lot of people mentally price their short liquidation as a mirror of the long and it is off by two maintenance margins.
Cross margin changes what is at stake
On cross margin, the whole account balance backs every open position. This pushes your liquidation price further away, which sounds like a gift and is really a trap. Your single position now has a much deeper buffer because your entire wallet is collateral, so the liquidation price on that position moves toward a level you might think you will never see. The problem is that when it does hit, it does not take out one position, it can cascade through everything at once, because they all shared the same collateral pool. I use isolated margin when I want a hard, known ceiling on what a single bad trade can cost me, and cross only when I am running a hedged book where the positions genuinely offset.
The cross calculation is the same shape as isolated, you just replace the position margin with total account equity minus the margin locked by your other positions. It is more annoying to compute by hand because it moves every time any position in the account moves. That alone is a reason to prefer isolated for directional bets you are trying to reason about cleanly.
Why maintenance margin tiers pull liquidation closer
Here is the part that catches experienced traders, not just beginners. That maintenance margin fraction is not a constant. Exchanges use a tiered schedule, and the tier is set by your position size in notional terms. Small positions sit in the base tier with a low maintenance requirement. As your notional grows, you cross into higher tiers where the maintenance margin fraction steps up, and on many venues the maximum leverage they will even allow steps down.
The practical effect is that a large position at nominal 20x might be forced into a tier whose maintenance requirement is several times the base, and sometimes the exchange will not let you hold 20x at that size at all. Your effective liquidation distance shrinks as the position gets bigger, even though nothing about the leverage number on your screen changed. Two traders both showing 20x, one small and one large, do not have the same liquidation price relative to entry. The large one is closer, sometimes meaningfully.
So the workflow before entry looks like this.
- Find the exchange's maintenance margin tier table for the specific contract. Every serious venue publishes one.
- Locate the tier your intended notional falls into, not the base tier you assumed.
- Use that tier's maintenance margin fraction and its max leverage in the formula above.
- Compute the liquidation price with those real numbers, then decide if the trade still fits your risk.
If the platform shows you an estimated liquidation price in the order ticket, good, but treat it as a starting point and sanity check it against your own number. Those estimates usually assume the fee and funding drag are zero, and both of those quietly push liquidation closer over the life of the trade.
What actually happens at liquidation
People imagine liquidation as the exchange closing their position at the liquidation price. It rarely works that cleanly. When your margin ratio hits the maintenance threshold, the liquidation engine takes over the position and tries to close it in the market. If it closes at a better price than the bankruptcy price, the leftover typically goes to the insurance fund. If the market is moving fast and it closes worse than the bankruptcy price, the insurance fund absorbs the shortfall so the counterparty still gets paid.
When the insurance fund itself cannot cover the gap, which happens in violent moves, some venues fall back to auto-deleveraging. That means the exchange forcibly closes part of a profitable trader's opposing position at the bankruptcy price to make the books balance. If you are on the winning side of a big move, you can get partially closed out through no fault of your own. It is rare, but it is a real mechanism, and it is the reason a stop order is not fully equivalent to just letting liquidation handle it. You also pay a liquidation fee on top, which is worse than a normal exit.
How much buffer to leave
Because liquidation is messy and expensive, I never want price to reach it. The stop is supposed to fire first, at a level I chose, with slippage I can estimate. The rule of thumb I use is to keep the liquidation price at least a couple of multiples of my expected slippage away from my stop, so that a fast wick through the stop still does not reach the liquidation engine before my exit fills.
Concretely, if my stop is 3 percent below entry, I do not want liquidation sitting at 3.5 percent below entry, because one bad candle closes that gap. I want it far enough that the stop has room to work even in a thin, gapping market. In practice that usually means running lower leverage than the venue permits, sizing so the position stays in the base maintenance tier, and computing the real liquidation price before entry rather than reading it off the ticket after. Do the arithmetic first and the trade either fits or it does not, either way not something you find out mid-wick.