Isolated Margin vs Cross Margin: What Each Mode Puts on the Line
Isolated margin and cross margin are the two ways a crypto exchange decides how much of your money stands behind a leveraged position. Isolated caps the loss at the margin assigned to that one position; cross lets the whole balance absorb losses, so liquidation comes later but can take everything.
Every leveraged crypto order form carries a small switch next to the leverage slider, and most traders leave it wherever the exchange put it. That switch decides whether a bad trade can cost the margin you posted for it or the entire balance in the account.
What each one is
Isolated margin gives one position its own pot of money. You assign an amount to the trade, and only that amount, plus the position’s own profit or loss, stands between it and a forced close. OKX’s maintenance margin rules put it in one line: isolated margin calculates requirements for each position separately.
Cross margin makes the account balance the pot. Every cross position draws on the same equity, so a loss on one is covered by money that was sitting idle, or by gains on another. The same OKX rules say cross margin calculates requirements from the combined risk across all positions. What a forced close involves, step by step, is on the crypto liquidation page; here the question is how each mode moves it.
Where they differ
What a single liquidation can take. In isolated mode the worst case for one position is roughly the margin assigned to it. In cross mode the worst case is the balance of the account, because that whole balance was counted as support.
How far the price can move first. More money behind a position means lower effective leverage, so the liquidation price sits further away in cross than in isolated for the same position size.
Whether positions affect each other. Isolated positions are sealed off from one another. Under cross, OKX counts the positions in all directions of a contract together for tier and maintenance purposes, whereas in isolated mode each direction of each contract is counted separately.
What happens to open orders. OKX’s isolated margin page, written for its margin-trading positions, says that before a liquidation the system cancels open orders sharing the at-risk position’s margin currency, and leaves the others alone. OKX’s tiered rules add that once an account’s ratio reaches 100% or lower, new orders that need more margin are canceled or rejected first; in cross mode that ratio covers the whole account.
How the exchange measures each one
Both modes use the same test: a maintenance margin ratio. OKX’s futures liquidation page defines it as the equity backing the position divided by the maintenance requirement plus a fee term. At 100% or below, the system starts cutting the position, tier by tier, and liquidates it fully if that is not enough. OKX says the liquidation price is based on the mark price, which its charts show as a price type separate from the last trade.
The difference is only in the numerator. In isolated mode it is the position’s own assets and margin against its own debt. In single-currency cross mode it is the cross balance plus unrealized profit and loss, less money reserved for open orders and less the margin already committed to isolated positions. So money placed in an isolated position is not available to rescue a cross position, and the reverse is also true.
Kraken describes a pooled design in its own terms. Its specification for multi-collateral perpetuals settles in US dollars but accepts eleven collateral currencies, bitcoin and euros among them, each counted after a haircut. It sets maintenance margin at half of initial margin, with initial margin as low as 2%. That page does not use the words isolated or cross, so this page does not map Kraken’s product onto either label.
A worked example
All figures here are hypothetical round numbers. An account holds $5,000. It opens one long position: 0.2 bitcoin at $50,000, a $10,000 position. Maintenance is 0.4% of the position’s value, OKX’s rate for its smallest BTC-USDT perpetual tier, and fees and funding are left out.
Isolated, with $1,000 assigned (10 times leverage). The position is closed where its equity equals maintenance: $1,000 + 0.2 x (price - $50,000) = 0.004 x 0.2 x price. That solves to $45,180.72, a fall of 9.64%. At that price $36.14 of maintenance margin remains, so the loss is $963.86 and the other $4,036.14 in the account is untouched.
Cross, with all $5,000 behind it (2 times leverage). Put $5,000 in place of $1,000 and the same equation gives $25,100.40, a fall of 49.80%. The loss at that price would be $4,979.92.
Now a 12% fall to $44,000. The isolated position is already gone. The cross position shows a loss of $1,200, leaving equity of $3,800 against maintenance of $35.20, a ratio of about 10,795%, nowhere near the 100% line. At a 30% fall, $35,000, the cross loss is $3,000 and the position is still open. At 50%, $25,000, cross has been liquidated for $4,979.92, while the isolated loss never grew past $963.86.
Which one to use
Use isolated margin when the trade is a single leveraged idea. The margin you assign becomes the maximum loss, and that number can be chosen with position sizing before the order goes in, rather than discovered afterward.
Use cross margin when you want the whole balance behind a position on purpose, for example when holding positions that offset each other, and only when the total position size divided by the whole balance stays low. The example above is 2 times leverage on the account in cross, which is why it took a 49.80% fall to end it.
Stay out of cross when one position is large relative to the account. A 50 times position backed by an account five times its margin is still 10 times leverage on everything you hold.
What stays the same in both modes
The maintenance rate itself. OKX’s public API returned identical tier tables for isolated and cross BTC-USDT perpetuals on 26 Sep 2026: 99 tiers, the first covering up to 10 bitcoin at a 0.4% maintenance rate and a 100 times maximum.
The fees and the mark price. OKX’s pages apply the same liquidation fee, charged at the taker rate, and the same mark-price test in both modes. A funding rate payment is paid or received either way; OKX’s funding page says the mode only changes whether it moves through the position’s isolated margin or the cross account’s equity.
The availability question. OKX’s pages state that not all products are available in all jurisdictions, and Kraken’s says geographic restrictions may apply, so whether either mode can be used at all depends on where the account holder lives.
The original data
The rule, fixed before counting: every bitcoin daily close is a new opening, judged on the next day’s bar alone. A long counts as liquidated if that day’s low reached its liquidation price; a short, if the high did. The positions are $10,000, $20,000 and $50,000, each with $1,000 in isolated mode, then again in cross mode with a $5,000 account behind them. Every opening uses OKX’s first-tier 0.4% maintenance rate, with no fees or funding, and the prices are Yahoo Finance’s BTC-USD daily bars for 4,391 days from 18 Sep 2014 to 25 Sep 2026.
Longs. The $10,000 position (10 times, isolated) was wiped out within one day on 110 of the 4,391 days, 2.51%; with the $5,000 account behind it in cross, never. At $20,000 it was 512 days (11.66%) isolated against 4 in cross. At $50,000 it was 1,804 days (41.08%) isolated against 110 in cross, the same count as the 10 times isolated position, because both are 10 times the money standing behind them.
Shorts behaved alike: 111, 547 and 1,912 days isolated against 0, 3 and 111 in cross. In the last three years alone, 1,096 days from 26 Sep 2023, the isolated longs were hit on 6, 77 and 427 days.
About half of those isolated liquidations were wicks. On 57 of the 110 days that ended the 10 times isolated long, bitcoin closed back above the liquidation price, so the cross version finished that day still open and down less than the isolated margin.
The other side of the trade-off is the worst day in the record. On 12 Mar 2020 bitcoin fell from a $7,911.43 close to a $4,860.35 low, 38.6%. Every isolated long was closed, losing $963.86, $923.69 and $803.21 before fees. The $10,000 cross long survived but was down $3,716.95 at the close. The $20,000 and $50,000 cross longs were liquidated, losing $4,939.76 and $4,819.28 of the $5,000.
The files. Both tables are in the one-day CSV and the 12 March 2020 CSV. The limits: each day is a separate hypothetical opening rather than a trading record, one exchange’s maintenance rate, daily bars rather than the mark price, and no fees or funding.
One simplification matters in the early years. The 0.4% rate is OKX’s tier for positions up to 10 bitcoin. While bitcoin closed under $1,000, $2,000 or $5,000, the $10,000, $20,000 and $50,000 positions were larger than that, so OKX’s own tiers would have charged more; the last such day was 12 Mar 2020. Using each day’s correct tier, the isolated long counts become 111, 521 and 1,907, and the $50,000 cross long 113, while the three-year counts do not change. On 186 days the $50,000 position was so large that OKX’s tier for it caps leverage at 40 times, so the 50 times version, still counted in the 1,907, could not have been opened at all.
On video, leverage draws a crowd and the mode does not. Of 24,971 trading videos in this site’s title study, 11 pair leverage with crypto or bitcoin, from 10 channels, at a median of 36,446 views and up to 701,276. The titles explain leverage or pitch it; the switch that decides what a liquidation can reach stays on the order form.
When it fails
Isolated fails when the margin is set too thin. A small allocation at high leverage is closed by ordinary intraday noise, and on half the days that ended the 10 times isolated long in the test, the price was back above the liquidation line by the close.
Cross fails when one position quietly becomes the whole account. A trader who opens cross at 50 times on a small idea has not risked a small amount; every dollar in the balance is now margin for it, as 12 Mar 2020 shows.
Both fail when the liquidation price is read from another exchange. Kraken sets maintenance at half of initial margin while OKX’s first tier is 0.4%, so the same leverage lands the line in different places. A stop-loss placed beyond that line never gets the chance to trigger.
And switching modes fails as a fix for a losing trade. Putting more of the account behind a losing position, by adding margin or by trading it in cross, only moves the line further away by putting more money in front of it.
Related
Crypto liquidation sets out the formula behind the lines drawn here and runs 30-day windows through it. The leverage page shows how a borrowed multiple enlarges gains and losses alike, and crypto futures describes the perpetual contracts on which traders usually meet this switch. For the stock-market version of the same choice, see cash account vs margin account.
I pick isolated mode before I pick the leverage, and I size the margin as the most I will accept losing on that one idea. Cross only makes sense to me once I have worked out the leverage of the whole account, not just of the trade.
— Michael Whitman
This page is educational, not financial advice. Test every idea on your own charts before risking money.