Crypto Liquidation: The Arithmetic, and 12 Years of Bitcoin Prices Run Through It
A crypto liquidation is the forced closing of a leveraged position by the exchange once the position's remaining margin falls to the maintenance level. The price where that happens is set by the leverage chosen, so it is known before the trade is opened.
A liquidation is not a surprise event, even though it usually feels like one. The price at which it happens is fixed by arithmetic the moment a leveraged position opens, and most exchanges display it. This page shows that arithmetic, checks it against an exchange’s own published example, then runs twelve years of bitcoin prices through it to see how often each level of leverage would have been hit.
How it works
A leveraged position is a loan plus a deposit. The trader puts up margin, the exchange extends the rest, and the position is sized on the total. The CFTC’s advisory on virtual currency trading puts the consequence plainly: profits and losses from price swings “are amplified in margined futures contracts”.
The exchange watches the margin that is left. Every loss comes out of the margin first. The exchange sets a floor, the maintenance margin, as a percentage of the position’s value. On OKX’s BTC-USDT perpetual, the smallest tier (up to 10 bitcoin) carries a maintenance rate of 0.4%, per the exchange’s API on 25 Sep 2026. Larger positions sit in higher tiers with higher rates.
What the exchange does at the line
When the margin falls to that floor, the exchange closes the position. OKX’s help page describes a tiered process: the system cancels open orders, reduces the position, and continues until it is fully closed if the account still falls short. It measures this against the mark price, a reference price the exchange computes, rather than against the last trade.
It happens at whatever price the market offers. A closing order during a fast move can fill worse than the liquidation price itself. OKX’s page says a liquidation fee applies and that extra costs may be charged to cover slippage, with any net amount going to its insurance fund.
The formula
For a long position, ignoring fees and funding:
Liquidation price = entry price x (1 - 1 / leverage) / (1 - maintenance rate)
For a short:
Liquidation price = entry price x (1 + 1 / leverage) / (1 + maintenance rate)
The first term is how far the price can move before the margin is gone. At 10 times, 1 / 10 = 10%. The second term moves the line slightly closer, because the exchange closes the position while a sliver of margin, the maintenance amount, is still there. With OKX’s 0.4% rate, a long at 2, 3, 5, 10, 20, 50 and 100 times is closed 49.80%, 33.07%, 19.68%, 9.64%, 4.62%, 1.61% and 0.60% below where it opened.
Checking it against the exchange’s own example. OKX’s help page gives a BTC-USDT perpetual long of 0.5 bitcoin at $60,996 in cross margin mode, with a 3 times leverage setting and $5,000 of available margin (the contract is priced in USDT, a dollar stablecoin), and shows an estimated liquidation price of 51,226.5 USDT. The result only reproduces with the whole $5,000 standing behind the position, which makes the $30,498 position about 6.1 times its margin rather than 3 times. At that leverage the formula above, with the 0.4% maintenance rate, gives about $51,201. Adding a 0.05% fee term to the maintenance rate gives $51,226.52, which rounds to the page’s 51,226.5. So the published figure is this same arithmetic with a closing fee built in.
A worked example
A hypothetical position: $1,000 of margin, 10 times leverage, long bitcoin opened at a hypothetical $100,000.
- Position size: $1,000 x 10 = $10,000, which buys 0.1 bitcoin.
- Liquidation price: $100,000 x (1 - 0.1) / (1 - 0.004) = $90,361.45, a fall of 9.64%.
- Loss at that price: 0.1 x ($100,000 - $90,361.45) = $963.86.
- What remains: $1,000 - $963.86 = $36.14, which is exactly 0.4% of the position’s value at $90,361.45. That remainder is the maintenance margin, and the fee and closing costs come out of it.
The same position as a short would be closed at $109,561.75, a rise of 9.56%.
A 9.64% fall is not a rare event for bitcoin. On 110 of the 4,390 trading days since 18 Sep 2014, the day’s low sat at least 9.64% below the previous day’s close. A 10 times long opened at the earlier close would have been gone within a day on each of them.
The original data
The test: open a position at every daily close of bitcoin from 17 Sep 2014 to 25 Aug 2026, hold it for 30 days, and ask whether the price reached the liquidation level at any point. Longs are checked against each day’s low, shorts against each day’s high, using Yahoo Finance’s BTC-USD daily bars. The maintenance rate is OKX’s 0.4% tier; fees and funding are left out, which makes these numbers conservative. That is 4,361 separate 30-day windows at each leverage level.
How often each level was reached, full period (long / short): 2 times, 0.9% / 7.6%. 3 times, 5.8% / 16.2%. 5 times, 21.0% / 32.4%. 10 times, 43.1% / 57.9%. 20 times, 64.5% / 77.4%. 50 times, 83.3% / 89.9%. 100 times, 91.4% / 94.9%.
The last three years only (1,066 windows opened from 25 Sep 2023, long / short): 2 times, 0.0% / 1.1%. 3 times, 1.3% / 6.9%. 5 times, 12.9% / 18.7%. 10 times, 37.6% / 51.9%. 20 times, 63.9% / 76.5%. 50 times, 84.6% / 90.7%. 100 times, 93.5% / 96.5%.
What the table says. Low leverage survived almost every month in the record, because bitcoin rarely fell by a third or a half inside 30 days. From 10 times up, liquidation was closer to the usual outcome than the exception. Shorts were hit more often than longs at every level, because over this period the price rose more often than it fell; that is a description of these twelve years, not a rule. The worst single day in the series was 12 Mar 2020, when the low sat 38.6% below the previous close, enough to close a 3 times long in one session. The windows overlap, so they are not independent trials; read the shares as how often the path went there, not as odds. The full table is in a CSV of the 30-day windows.
Market-wide liquidation totals are not used here. No source found for this page publishes a total across all exchanges that could be checked back to the exchanges themselves, so none is quoted.
When it fails
When the stop sits beyond the liquidation price. A stop-loss that would trigger after the exchange has already closed the position does nothing. At 20 times, the liquidation line is 4.62% away.
When a daily chart hides the move. Liquidation is checked continuously against the mark price. A wick that lasts minutes is enough, which is why this test uses daily lows and highs, not closes.
When the position grows into a higher tier. Maintenance rates rise with size, so a bigger position at the same leverage is closed sooner.
When funding keeps draining the margin. A funding rate debit comes out of the same margin, nudging the liquidation price closer the longer a paying position is held.
When many positions share one level. Each forced close is a market order in the same direction. That is how one liquidation can move the price into the next, the cascade the crypto futures page describes.
Related
Leverage covers what borrowing does to both sides of a result, and position sizing is the method for choosing a size from the risk rather than from the multiple on offer. The margin call calculator runs the same kind of arithmetic for a broker’s margin account, and crypto futures explains the perpetual contracts where most crypto liquidations happen.
I pick the liquidation price first and let it set the leverage, not the other way round. If the level where the exchange closes me out sits inside an ordinary week’s range, the position is too big for the account.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.