Leverage vs liq price, the math

1. What leverage actually is — notional vs margin

You have 1,000 USDT in your OKX account, you pick 10x leverage, and you open a BTC-USDT-SWAP long. The system gives you a 10,000 USDT notional position, and 1,000 USDT of that sits as your initial margin.

PnL on the position calculates against the full 10,000 USDT notional, while your downside collateral is the 1,000 USDT only. BTC up 1% means you gain 100 USDT, which is 10% of margin. BTC down 1% means you lose the same 100 USDT, again 10% of margin. That is the entire magic of leverage.

The interesting question is when do you actually get liquidated. Naive answer: 10% adverse move (because margin ÷ leverage = 10%). Half right. The real liq price is closer than that.

2. Isolated vs cross — two different risk models

OKX perps support two margin modes — Isolated and Cross. Pick the wrong one without knowing the difference and you can hand back the entire account on a single misread.

Isolated. The margin allocated to this position is walled off from the rest of your balance. If the position liquidates, you lose that walled-off chunk and nothing else. Account has 5,000 USDT, you assign 1,000 to a BTC long, liquidation costs you 1,000, the other 4,000 is intact.

Cross. Every available USDT in the account is available to defend this position. It can absorb a bigger drawdown, but the failure mode is the entire account — 5,000 USDT in, 0 out.

Our desk's hard rule: the first six months are isolated only. Cross mode is for traders running offsetting books (e.g. long BTC + short ETH where the combined delta is near zero). Even then, the first time you touch a new perp, you size it isolated.

OKX defaults to isolated when you open a new position, but plenty of traders flip to cross by accident. Before you do, refresh the margin fundamentals on Investopedia's Margin entry — initial vs maintenance margin, the maintenance ratio, and what liquidation actually means — then compare the two modes side by side using your actual numbers.

3. Maintenance margin rate (MMR) — the variable that matters

Maintenance margin rate is the percentage at which your position margin ratio is forced to liquidate. On BTC-USDT-SWAP, MMR ranges from about 0.4% (position notional under 5M USDT) to 0.6% (5M-10M USDT).

The definitions stack up cleanly. Position margin ratio = account margin balance ÷ notional. At the moment you open, your margin ratio equals 1 / leverage — so a 10x position starts with a 10% ratio. As the price moves against you, the loss eats your margin, and the ratio falls. When it hits MMR (0.4% or 0.6%), the system liquidates.

For the textbook framing of MMR, Investopedia's maintenance margin entry drops it into the traditional futures context. For where the actual liq clusters live, CoinGlass liquidation data shows the real distribution across pairs and venues.

4. Worked liq price: 1,000 USDT × 10x BTC long

Walk through with numbers. Margin 1,000 USDT, 10x leverage, isolated, BTC-USDT-SWAP long at $70,000. Notional = 1,000 × 10 = 10,000 USDT, position size = 10,000 ÷ 70,000 ≈ 0.143 BTC.

Liq price formula (long, isolated): Liq Price = Entry × (1 − 1/Leverage + MMR + Fee Buffer). Take MMR at 0.5% and a fee buffer of 0.05% (the closing taker fee).

Plug in: Liq Price = 70,000 × (1 − 0.1 + 0.005 + 0.0005) = 70,000 × 0.9055 = $63,385. BTC dropping from $70,000 to $63,385 (a 9.45% move) wipes your 1,000 USDT.

That is 0.55% earlier than the "10% down and you blow up" intuition. Looks tiny on paper. On BTC's daily range it is one ordinary candle. Build that 0.55% into your mental stop, and you skip a lot of "I did the math and still got liquidated" moments.

Shorts get the symmetric formula: Liq Price = Entry × (1 + 1/Leverage − MMR − Fee Buffer). A 10x short at $70,000 liquidates at $76,615 — a 9.45% rip.

5. Comparison table: 5x vs 10x vs 50x

Run the formula at three leverage points. BTC long at $70,000, 1,000 USDT margin, isolated, MMR 0.5%.

• 5x → liq price $56,385, allowed drawdown −19.45%
• 10x → liq price $63,385, allowed drawdown −9.45%
• 20x → liq price $66,885, allowed drawdown −4.45%
• 50x → liq price $68,985, allowed drawdown −1.45%
• 100x → liq price $69,685, allowed drawdown −0.45%

The 5x vs 50x contrast is where the leverage decision actually lives. 5x gives you 19.45% of room — enough to survive a one-day shock like the 2024-08-05 flash crash (BTC −19.6% on the day, a 5x long would have liquidated almost exactly at the bottom). 50x gives you 1.45%. BTC routinely moves that much in an hour, no news required.

The 100x row is the giveaway. −0.45% allowed drawdown means random price wiggle plus your own opening fee can liquidate you. 100x on BTC perp is a psychological bait, not a serious trading instrument.

6. Partial liquidation — OKX's "trim first" mechanism

Plenty of traders think liquidation is binary — price hits level X and the position vanishes. OKX does not work that way. When your position margin ratio approaches MMR, the engine first trims 25-50% of the size to push the ratio back above the threshold. You keep some of the position and you keep trading.

That trim helps, in calm tape. On the 2024-08-05 flash crash, where BTC dropped $1,500 in 10 seconds, the partial-liquidation market orders printed at the very worst price of the move. In retrospect "you got half your position closed at the bottom" is the correct description.

The partial-liq trigger sits around margin ratio = MMR + 1%. On a 10x BTC long, that is the price level where your ratio falls from 10% to about 1.5% — call it 8.5% adverse move. That is 0.5% earlier than the $63,385 math liq price.

Working rule: do not trust the "1% cushion" mental model. The trigger for the first partial liquidation in production runs 0.5-1% earlier than the algebraic liq price. Build that buffer into where you place your stop.

7. 2024-03-14 BTC ATH and a 12.3B USD OI snapshot

On 2024-03-14 BTC printed its then-all-time high at $73,800 on Coinbase. OKX BTC-USDT-SWAP open interest crossed $12.3B that day, one of the highest readings of the trailing 12 months. The leverage structure inside that OI tells you something useful.

We pulled a snapshot of /api/v5/public/open-interest?instType=SWAP&uly=BTC-USDT at the time. Total OI $12.34B notional. Long-side OI ≈ $6.89B, short-side ≈ $5.45B. (Long and short notional must net out on perps; the apparent imbalance comes from delta-neutral hedges accounting in different buckets.) Long/short ratio: 1.26. Structurally bullish positioning.

Funding locked in at +0.034% per 8h that day — near the upper cap. Annualised that is +37.2%. On a 10x long, holding it for a year would have cost you 37% of margin just in funding. That is the carry profile in a high-OI, high-funding regime.

Over the next two weeks BTC retraced from $73,800 to $61,000, a 17.3% drop. OKX long-side liquidations cumulated to about $890M. A 10x long opened at the ATH on 2024-03-14 had a liq price of $66,793 — which printed on 2024-03-20.

8. What leverage should a new trader actually use

After Sections 1-7 you can compute liq prices yourself, you understand partial liquidation, and you know the real distance between 5x and 50x. The remaining question: what should you actually run?

Our recommendation is 3x-5x. That gives you headroom for a 10% adverse single-session move without going to zero, and it preserves enough magnification that a correct call still pays for itself.

10x and above are for traders with a documented win rate above 60% and tight stop discipline, plus at least 30 days of records. A new trader can run 10x if you pair it with a hard 1.5% stop — that way the worst case is still only 15% of margin gone.

20x and up — 50x, 100x — is negative expected value for 95% of perp participants. The leverage exists because it generates fee revenue for the platform. At 50x, the round-trip fee and funding eat 30-50% of any winning move. That is not a moral judgment, it is a math fact.

Next, read the stop-loss / position-size paradox to fold leverage and stop placement into the same equation. Or open the position calculator and run your next planned trade through it.