How Hyperliquid liquidations actually work (and why they surprise you)

Updated 2026-04-14

Most traders coming to Hyperliquid from Binance or Bybit underestimate how different the liquidation surface is. The math looks familiar, but the failure modes — when and why the number you computed in your head turns out to be wrong — are specific to Hyperliquid’s architecture. This guide walks through the full mechanism and the three places it bites.

The core formula

For an isolated long position with no funding accrual and no fees, Hyperliquid liquidates when:

accountEquity < positionSize × MMR

Where MMR is the maintenance margin rate from Hyperliquid’s tier schedule. At small notionals it’s 1%; it rises with size. Solve for price and you get the formula the calculator uses on this page. Try it:

Where the formula misleads you

1. MMR is tier-based, not flat

Most traders memorize “maintenance margin is 1% at 50x leverage” and stop there. On Hyperliquid, MMR steps up at discrete notional thresholds. A $99,000 long at 10x has different MMR than a $101,000 long at 10x — and the crossover can shift liquidation by 50-100 bps. This is identical in spirit to Binance’s tier schedule, but the step sizes are different; copying your Binance mental model will mislead.

The calculator above models this correctly — drag the position size across $100,000 and you’ll see the liquidation price jump.

2. Funding accrues hourly, not every 8 hours

Every CEX you’ve used settles funding every 8 hours. Hyperliquid settles every hour. Over a 24-hour hold that’s 24 funding charges vs. 3 on Binance. If the position is against a heavy funding rate, this eats your margin 8× faster than you’re used to. The liquidation price itself doesn’t encode funding — the calculator models the initial liquidation. A week-held position against funding will be liquidated sooner than the static formula suggests.

Practical rule: if funding is > 0.01% (annualized 87%), don’t hold overnight unless the thesis explicitly prices in the cost.

3. The oracle, not the mark, triggers liquidation

Hyperliquid uses a median of multiple external oracle feeds for liquidation triggering. This is usually boring. During acute volatility — the kind where CEX mark prices spike and immediately recover — the Hyperliquid oracle can lag 200-500ms behind, which is enough to wipe positions that would have survived on Binance. This is the opposite failure mode of what you’d expect; you might think a “slower” oracle would be friendlier, but a lagging oracle during a sharp down-spike can print a fill far below the actual cross-venue market.

4. Insurance fund and auto-deleveraging

When a liquidation doesn’t fully clear at the maintenance price (the system can’t find liquidity at the right level), Hyperliquid’s insurance fund absorbs the slippage. If the fund is depleted, auto-deleveraging (ADL) kicks in: the system closes profitable counter-positions to cover the loss. If you’re the trader holding a well-placed profitable short during a cascade, you may find your position partially closed by ADL before you intended to exit. The calculator cannot predict this; it’s a structural risk of using any DEX perp with socialized loss.

Isolated vs cross margin on Hyperliquid

Isolated: your liquidation price depends only on the margin you assigned to the position. Clean, predictable, and the default for position-level risk budgeting.

Cross: the liquidation price is computed against your total account equity, so unrelated positions and idle USDC push the liquidation far away. This is powerful but makes one bad decision capable of wiping the account — we’ve seen experienced Binance traders get ADL’d on Hyperliquid because they ran cross margin and forgot how much it widens the blast radius.

The calculator supports both modes. Set “Margin mode: Cross” and supply your total account equity.

A concrete example

Say you open a $50,000 notional long on ETH-PERP at $3,000 entry, 10× isolated leverage, fees excluded.

• Initial margin: $5,000.
• Tier 1 MMR at $50k: 1.0%.
• Maintenance requirement: $500.
• Buffer before liquidation: $5,000 − $500 = $4,500.
• Price move that eats the buffer: ($4,500 / $50,000) × $3,000 = $270.
• Liquidation price: $3,000 − $270 = $2,730 (−9.0% from entry).

Now include fees (0.035% open + 0.035% close = 0.07% round-trip on $50,000 = $35):

• Adjusted buffer: $4,500 − $35 = $4,465.
• Price move: $267.90.
• Liquidation price: $2,732.10 (−8.93% from entry).

Tiny change from fees at this size, but it compounds noticeably at higher leverage or larger positions.

Sizing and safety margins

If your stop-loss is within, say, 20 bps of the computed liquidation price, you’re taking real liquidation risk. Two things can push you through the formula: oracle lag during a volatile minute (point 3 above) and funding accrual if the hold is long.

Practical rule: require a 2× cushion between stop and liquidation. If you want to stop at −5%, the liquidation should be ≥ −10%. The calculator makes this easy: solve for the leverage that gives you that cushion.

What the calculator doesn’t model

• Funding accrual over the hold. Initial liquidation only.
• ADL risk. Structural, not computable from position inputs alone.
• Oracle lag. Assumes mark = oracle.
• Insurance fund socialization. Not modelled.

For any of those, rely on Hyperliquid’s own risk dashboards and your own position management discipline.

Related tools

• Hyperliquid Liquidation Calculator — the calculator used above.
• More venues and strategy calculators ship on the roadmap; sign up below if you want to know when.

Last verified against Hyperliquid’s live UI on 2026-04-14. If Hyperliquid ships parameter changes we haven’t updated, send me a DM on X (see about).