Crypto Liquidation Price Formula: Long and Short Examples

Take a linear contract: each one-dollar price change produces a one-dollar profit or loss per coin of exposure. The liquidation calculation asks how far that price can move before the remaining support becomes too small. It does not ask when every dollar has disappeared. Maintenance margin is the amount that must still remain while the trade is open.

Four inputs you can recognize on a position ticket

Use the actual average entry, the filled coin quantity, the net margin assigned to this position, and its applicable maintenance rate. An order for 1 BTC that fills only 0.6 BTC is a 0.6 BTC position. An account with 6,000 USDT in total is not necessarily a position with 6,000 USDT assigned. Those two substitutions are enough to make an otherwise correct equation give a meaningless answer.

At a trial price P, the long has equity M + q × (P − E). Our maintenance requirement is q × P × m. Put those expressions equal and solve for P. The price appears on both sides because, in this model, the value on which maintenance is charged changes as the market changes.

If initial margin is exactly q × E ÷ leverage, you can substitute that expression for M. If you later add or remove collateral, use the resulting margin amount instead of pretending the opening leverage still describes it. The formula is most useful when every number corresponds to an observable field, rather than a remembered slider setting.

Both sides use the same settlement unit. For BTC/USDT, multiply BTC by USDT per BTC to obtain USDT. A BTC-margined inverse position has a different payoff and cannot be repaired by simply changing the currency label. The unit check is a quick way to catch a wrong calculator before relying on its result.

Substitute the example into the balance equation: 6,000 + P − 60,000 = 0.005 × P. Move the price-dependent maintenance to the left: 0.995 × P = 54,000. Dividing gives 54,271.3568. Rounding the displayed estimate to two decimals gives 54,271.36 USDT per BTC.

Check the answer by putting it back into both balances. The position has lost 5,728.6432 USDT, leaving 271.3568 of equity. Its value is now 54,271.3568, and half a percent of that is also 271.3568. The two sides meet. That reconciliation is more informative than a liquidation number shown without its supporting balances.

What the quick leverage shortcut leaves out

At 10×, one divided by leverage is 10%. Subtracting 10% from 60,000 gives 54,000, the zero-equity point in this simplified example. It cannot also be the maintenance boundary when a positive amount must remain. A more refined shortcut adds the maintenance percentage to the entry-price calculation and gives 54,300. That is close here, but still treats the maintenance base differently.

The 28.64 difference between 54,300 and 54,271.36 is small compared with Bitcoin’s price, yet it can matter if a protective order is placed nearby. More importantly, a close approximation in this one case does not validate it for another leverage, maintenance rate or margin system. Compare the assumptions before comparing decimal places.

Keep full precision during the calculation, then show the estimate at readable precision. A venue may round its display to a permitted tick while evaluating the underlying margin condition more precisely.

For the corresponding short, adverse movement is upward. Its equity is 6,000 + 60,000 − P. Set this equal to 0.005 × P and solve: 66,000 ÷ 1.005 = 65,671.64. Equity and maintenance meet at about 328.36 USDT. The short and long distances are not exactly symmetric because the short’s position value grows during the adverse move.

Now keep the quantity and entry fixed. Adding 1,000 USDT of margin moves the modeled long boundary down to 53,266.33; it moves the short boundary up to 66,666.67. No BTC has been bought or sold by that addition. You have placed more collateral behind the same price exposure, so the position can absorb a larger adverse cash result before reaching maintenance.

A debit matters only if it reduces the support being modeled

Suppose 30 USDT of paid funding is deducted from this isolated allocation. Support falls from 6,000 to 5,970. If a separate fixed 20 USDT must also be held for closing costs, only 5,950 remains available before the maintenance test. Our long result becomes 54,321.61 and the short result becomes 65,621.89. Both move toward entry.

The simulator treats those adjustments as a single fixed deduction so the relationship stays visible. Real venues may debit funding from a different balance or calculate a closing reserve from a price-dependent formula. In that case, enter the amount actually affecting the position and use the venue’s rule for the reserve. Do not subtract the same fee once from margin and again as an extra adjustment.

Receiving funding has the opposite effect only when the credit contributes to the support used in the test. A credit sitting elsewhere in an account need not rescue an isolated position. The account ledger tells you where the cash went; the funding rate alone does not tell you which liquidation estimate should move.

A larger position may enter a higher maintenance tier. The applicable rate and any tier deduction must be used together. The deduction offsets maintenance already accounted for in earlier bands; it is not cash credited to the wallet or a discount that you can withdraw.

If the candidate price falls outside the tier used in the equation, solve again with the tier that contains the resulting value. For shorts, rising prices can move a position into a higher band without adding contracts.

Bybit values the position at mark and can include estimated closing fees in the displayed requirement. Separate those fees and any open-order reservations before comparing that panel with a formula for one filled position.

Start with the largest possible mismatch: the product and margin mode. A cross account, an inverse payoff or a multi-asset wallet is not the isolated linear example above. Once those match, work through the inputs in the order that money actually entered and left the position.

For a partially filled order, use the fill average rather than the limit you originally requested. If you add to an existing position at a different price, both the quantity and average entry change. Replacing just the quantity while retaining the old average creates a discrepancy that looks like a margin-rule issue but is really a fill-history issue.

Bybit’s current UTA help page publishes separate isolated formulas for USDT, USDC and inverse contracts. Its USDC section also explains how session settlement updates the average entry and position accounting. A reset entry can therefore be legitimate. Compare the post-settlement balances rather than combining today’s entry with yesterday’s unrealized result.

Keep a compact record of the inputs at the same moment: contract name, mode, quantity, average entry, assigned support, tier and displayed estimate. Comparing a screenshot from before a funding payment with a wallet balance after it can manufacture a contradiction. A useful record reconciles one state of the account; it does not need a folder full of unrelated charts.

If you still cannot reconcile the result, the unresolved item should be specific: for example, whether the closing reserve is already included in maintenance. That is a question the product documentation or support can answer. “The formula is wrong” is much less actionable than identifying the exact balance or convention that differs.

A cross account combines several moving values. To quote a single BTC liquidation price while an ETH position is also open, a calculator must decide what happens to ETH and every collateral asset along the way. Holding them constant is a scenario, not a prediction about a stressed market.

Consider an account with 2,000 USDT of equity and a combined maintenance requirement of 300. The current surplus is 1,700. If another position loses 800 while BTC stays unchanged, that surplus falls to 900 even before considering any change in maintenance. The BTC chart has not moved, but there is less shared support behind the trade.

Kraken’s account-wide test illustrates the collateral problem

Kraken Multi-M values non-USD collateral in dollars with haircuts. Its documentation distinguishes an isolated-position test, cross liquidation that preserves isolated allocations, and an account-wide test. A fall in collateral value can therefore matter even when a particular contract’s own price result is unchanged. Isolated funding payments are documented as coming from the general wallet balance, another reason not to apply one assumed funding path everywhere.

Imagine supporting an ETH contract with BTC collateral. An ETH-only calculation can stay constant while the dollar value of BTC falls. The missing variable is the support asset, not the ETH contract quantity. To understand the account boundary you need a collateral-price scenario as well as a contract-price scenario.

Hyperliquid similarly explains that an open position’s displayed estimate can change with funding and, for cross margin, unrealized results elsewhere. Its documented formula also depends on the margin tier at the liquidation price. This makes the estimate useful for the current account state, but not a fixed promise that survives later deposits, withdrawals or new trades.

After a partial close, recalculate with the remaining quantity and equity. Realized P&L, execution costs and a lower maintenance tier can all change the new boundary. Keeping the old entry price does not justify keeping the old liquidation estimate.

The boundary calculated above identifies a maintenance condition. It does not forecast the average price at which an exchange can close the position. The mark used for the risk test and the bids or offers available for execution serve different purposes.

In the simple long example, zero equity occurs at 54,000, whereas maintenance is reached around 54,271.36. Calling both values “the liquidation price” hides the buffer between them. An actual venue’s bankruptcy calculation can include additional accounting rules; use its definition when interpreting an account statement.

Hyperliquid describes an initial attempt to close through the order book, followed by backstop treatment if support deteriorates sufficiently. This distinction means the result of the process depends on execution and the remaining collateral, not just the first mark that crossed a threshold. It is one reason a liquidation record may contain several prices rather than one clean number.

A protective stop belongs earlier in the trading plan. Setting it at the displayed liquidation estimate gives the engine and the order little room to behave differently. The next guide examines order triggers and fills; changing the formula here cannot make an exit order guaranteed.