Crypto liquidation price calculator

A perpetual position starts with a notional value and a posted margin. The margin is the notional divided by the leverage, so a 1,000 dollar position at 10x rests on 100 dollars. As the price moves against the position the unrealised loss eats that margin. The venue does not wait for the margin to reach zero. It closes the position once the remaining equity falls to the maintenance requirement, which is a small percentage of the notional set by the venue and usually stepped by position size.

Setting the unrealised loss equal to the posted margin minus the maintenance requirement and solving for price gives the level below. The direction of the trade only flips the sign. The distance itself is the same either way, which is why a short at 10x sits as far above entry as a long at 10x sits below it.

Search this term and the tools that come back all return the same arithmetic. Enter an entry price and a leverage, receive a liquidation price and a distance in percent. That distance is correct. It is also almost useless on its own, because 9.1 percent means nothing until you know what the asset in front of you does in a week.

So this page puts the distance next to the record. For the selected token it shows the measured 24 hour, 7 day, 30 day, 200 day and 1 year changes, plus the current drawdown from the all-time high, and it counts how many of those readings are already at least as large as the liquidation distance you just created. When the answer is four out of five, the leverage setting has stopped being an abstraction. The move that closes the position is not hypothetical for that asset. It is in the record, and it happened inside a window shorter than most people hold for.

The count reads the absolute size of each change, so an upward move counts against a short and a downward move counts against a long, and a window with no reading is excluded rather than filled in. CoinGecko does not serve a 90 day window on this endpoint, so no 90 day figure appears anywhere on the page.

The second table widens the question from one token to the whole embedded set. For the leverage on screen it takes the liquidation distance and asks, across the 212 non stablecoin tokens in the snapshot, how many of them recorded a move at least that large over the chosen window. The count and the denominator are both printed, and the tokens the source gave no reading for are reported as an explicit exclusion rather than being quietly dropped from the base.

That share is a backward looking frequency over a single window ending on the fetch date. It says how common a move of that size was among these assets in that window. It is not a probability that a position opened today gets closed out, and nothing here should be read that way. The sample is one point in time and the assets in it are the ones large enough to sit near the top of the market today, which is its own selection effect.

The calculator also solves the question in reverse. Given the largest absolute move the selected token made across those windows, it reports the highest leverage whose liquidation distance still clears that move. That figure is frequently in the low single digits, and for a token that halved over a year it can fall below 1x, in which case the tool says so rather than rounding up to something reassuring.

A liquidation is a market order the position holder does not get to place. The engine has to sell into whatever book exists at that moment, and on a thin book that sale moves the price further, which reaches the next stop below, which sells again. That feedback is why liquidation clusters arrive as cascades rather than as single prints, and why the same leverage carries a different practical risk on a deep pair than on a thin one.

Early Thunder scores 251 tokens across 25 variables, and one of them is exchange depth. The panel joins the selected token to that file on its ticker and prints the composite score, the verdict and the exchange depth sub score against the universe median, with an explicit not scored state when the token has no row. The file was generated on 2026-09-18, so it is a dated research snapshot and the depth reading describes the venues as they were then. The full methodology is on the methodology page, and every scored token has its own page in the scorecard.

Matching is on the ticker symbol, the weakest join available in this market. Tickers are reused across chains and reassigned after rebrands, so a match is evidence and not proof. This build matched 109 of the 212 embedded tokens.

Funding is the first gap. A perpetual pays or collects funding every few hours, and on an isolated position that payment comes out of the same margin the liquidation level is computed from. A long paying funding through a flat market watches its liquidation price drift up towards the mark without the price having moved at all. This model takes a single snapshot and applies no funding, so it understates the level on a position held for days.

The mark price is the second. Venues do not liquidate on the last trade. They liquidate on a mark that is usually an index of several spot venues, sometimes smoothed by a moving average or a funding basis term. A wick on one exchange that never touches the index will not close the position, and an index that lags a genuine move will close it late. The level here is computed against the entry price you enter, which is a different quantity from the venue mark.

Fees and slippage are the third. The forced close pays a taker fee and often a separate liquidation penalty, and it executes wherever the book allows rather than at the computed level. On a thin pair the realised close is worse than the arithmetic, which is exactly when the depth reading matters most. Cross margin is the fourth. Once the account is cross, the whole balance stands behind every open position, so the real level depends on every other position you hold and moves whenever any of them does. The free balance field models one moment of that, not its path.

The maintenance rate is the fifth. Venues tier it by position size, and the rate here is whatever you type, applied to the entry notional rather than to the notional at the liquidation level. That is the common simplification and not the exact venue formula, so check the tier table for the contract before trusting a level from any calculator, this one included.