The exact liquidation price formula, derived and explained

Most tutorials put the liquidation price at entry times one minus one over leverage. It is a handy shortcut, but it leaves out the maintenance margin and shifts the result exactly where precision matters most: on highly leveraged positions. This guide derives the formula Binance uses on USDⓈ-M futures, compares it with the shortcut and explains how margin brackets and cross margin change the picture.

What a liquidation price is

Opening a leveraged position takes an initial margin equal to the position's value divided by leverage. The exchange also requires the margin to stay above a floor, the maintenance margin, set as a percentage of position value called the maintenance margin rate (MMR). The liquidation price is where initial margin plus unrealised profit or loss falls to that floor.

Two details matter. Binance checks against the mark price, not the last trade, so a wick in the order book is not enough if the mark price does not follow. And the formula below applies to an isolated-margin position in one-way mode with nothing else open, which is the case where the margin backing the position is exactly what you see.

The exact liquidation price formula

LP_long = [EP × (1 − 1/L) − cum/Size] / (1 − MMR)
LP_short = [EP × (1 + 1/L) + cum/Size] / (1 + MMR)
  • LP: liquidation price.
  • EP: entry price.
  • L: leverage.
  • MMR: maintenance margin rate of the bracket the position's notional falls into.
  • cum: cumulative maintenance amount of that bracket; zero in the first one.
  • Size: position size in contract units.

In the first bracket, which covers small positions, cum is zero and the formula collapses into a compact form:

LP_long = EP × (1 − 1/L) / (1 − MMR)
LP_short = EP × (1 + 1/L) / (1 + MMR)

Deriving the liquidation price step by step

For a long, the initial margin is Size × EP / L, the unrealised result at price LP is (LP − EP) × Size, and the maintenance margin at that price is Size × LP × MMR − cum. Liquidation happens when the first two add up to the third:

Size × EP / L + (LP − EP) × Size = Size × LP × MMR − cum
LP × Size × (1 − MMR) = Size × EP × (1 − 1/L) − cum
LP = [EP × (1 − 1/L) − cum/Size] / (1 − MMR)

The short case follows the same steps with the profit term flipped to (EP − LP) × Size, which is why its formula adds where the long one subtracts. The final division by (1 − MMR) or (1 + MMR) appears because maintenance margin is charged on the position's value at the liquidation price, not at entry. That is the piece the shortcut drops.

Shortcut vs exact formula: how far off the liquidation levels get

Take a 100x BTC long entered at 100,000 USDT with the 0.4% MMR of BTCUSDT's first bracket. The three common versions disagree:

Formula100x long LP from 100,000
Shortcut: EP × (1 − 1/L)99,000.00
Shortcut plus MMR: EP × (1 − 1/L + MMR)99,400.00
Exact: EP × (1 − 1/L) / (1 − MMR)99,397.59

The plain shortcut lands 397.59 USDT below the real level; adding the MMR overshoots by 2.41 USDT. That sounds small, but a short-timeframe liquidation heatmap works in 0.05% price cells, about 50 USDT with BTC at 100,000. The shortcut moves the level almost eight cells and draws it where nothing is. Here is the exact formula applied to four leverage levels on both sides:

Illustrative example with a round entry of 100,000 and an MMR of 0.4%. Not a current price or a recommendation.
LeverageLong liquidationDistanceShort liquidationDistance
10x90,361.45−9.64%109,561.75+9.56%
25x96,385.54−3.61%103,585.66+3.59%
50x98,393.57−1.61%101,593.63+1.59%
100x99,397.59−0.60%100,597.61+0.60%

Margin brackets: where MMR and cum come from

MMR is not one number per coin. For each contract Binance publishes a ladder of brackets by position notional: the larger the position, the higher the MMR and the lower the maximum leverage. BTCUSDT's first bracket carries a 0.4% MMR, ETHUSDT's 0.5%, and altcoins usually sit between 1% and 2.5%. Bybit and OKX publish equivalent ladders.

cum keeps the ladder continuous. Without it, maintenance margin would jump the moment a position crossed from one bracket into the next, so Binance subtracts a fixed amount per bracket. Dividing it by Size is why it weighs more on a position that only just made it into a higher bracket.

The bracket's effect is real but bounded. On BTC, moving from a 0.4% to a 0.5% MMR shifts the liquidation price by 0.1%. On altcoins with brackets from 1% to 2.5% the shift is larger. liqlin loads every bracket from the exchange itself rather than typing them in, and for each leverage level on the map it uses the bracket that holds that level's typical position size.

Liquidation price in isolated vs cross margin

In isolated margin a position only has the margin assigned to it, and the formula above holds. Topping up margin by hand, or paying and receiving funding, changes that margin and therefore the liquidation price. In cross margin the cushion is the whole account balance: the position survives longer, and its liquidation price depends on every other position and on free balance, which nobody outside the exchange can see.

This is not an edge case. A measurement on 2026-10-05 over 31,640 real Hyperliquid positions found 90.6% of notional in cross margin, and for that group the median real distance to liquidation was 5 to 12 times the distance implied by the declared leverage. So the liqlin map does not place money by the leverage each trader declares, but by the distance at which positions actually liquidate: an effective leverage.

Frequently asked questions

Why does my exchange show a slightly different liquidation price?

The formula assumes isolated margin, a single position and the right bracket. Fees, funding paid or received, added margin and cross margin all change the margin available and move the result.

Is liquidation triggered by the last price or the mark price?

On Binance, by the mark price. A move in the last traded price that does not drag the mark price along does not liquidate the position.

What happens if leverage is higher than 1/MMR?

The formula would put the liquidation price on the wrong side of the entry, which makes no sense. Exchanges prevent it by capping the maximum leverage of each bracket, and the liqlin map respects that cap.

Does the same formula apply to Bybit, OKX and Hyperliquid?

The structure is the same: initial margin plus unrealised result equals maintenance margin. What changes is each contract's bracket ladder and MMR, which liqlin reads from each exchange.

Related maps

Guides to read the map