Hyperliquid Liquidation: Formula & HYPE Tiers

Hyperliquid evaluates maintenance margin using the mark price rather than waiting for a trader's preferred exit. For an isolated position, adverse unrealized PnL reduces the margin assigned to that position. Once the remaining value reaches the maintenance requirement, the position can enter the liquidation process.

Liquidation price is therefore a margin boundary, not an order resting in the book. The actual path can include partial liquidation, book execution, and backstop handling. A calculator answers where the simplified position equation reaches maintenance; it does not quote the price at which every liquidation fill will occur.

For one isolated perp position, the useful equation is: isolated margin plus directional unrealized PnL equals maintenance margin at the liquidation price. Direction is positive for a long and negative for a short. Unrealized PnL is direction multiplied by the difference between liquidation price and entry price, then multiplied by base-asset size.

Tiered maintenance is not simply liquidation notional multiplied by one rate. Hyperliquid's published tier model uses maintenance margin rate and a maintenance deduction so the requirement remains continuous when notional crosses a tier boundary. HypeBasis solves the position equation against every valid tier, then keeps the solution whose liquidation notional falls inside that tier.

• Position value at liquidation: margin + direction × (liquidation price − entry price) × size.
• Maintenance at liquidation: liquidation notional × maintenance rate − tier deduction.
• Official maintenance rate: half the initial-margin rate implied by the tier's maximum leverage.

As reviewed on August 25, 2026, the official HYPE table exposes 10x maximum leverage below the $20 million notional boundary and 5x from that boundary upward. Those limits imply maintenance rates of 5% and 10%. The upper tier carries a $1 million deduction, which prevents a discontinuous jump in required maintenance at $20 million.

Consider a 300,000 HYPE short entered at $80 with $4.8 million of isolated margin. Entry notional is $24 million. The tested tier-aware equation estimates liquidation near $90.30, where notional is about $27.09 million, so the 5x tier, 10% maintenance rate, and $1 million deduction apply.

Now change the scenario to a 300,000 HYPE long at the same $80 entry with $6 million of isolated margin. The estimated liquidation price is about $63.16, or roughly $18.95 million of notional. Because price fell far enough to move the position below the tier boundary, the valid solution uses the lower 10x tier and 5% maintenance rate. Applying the entry tier forever would give the wrong answer.

For a long, a stop above the estimated liquidation price sits earlier on an adverse move. For a short, a stop below the estimated liquidation price sits earlier. If the order is on the other side of liquidation, the stated stop loss should not be used to calculate reward to risk because the modeled position reaches maintenance first.

Even a correctly placed stop does not guarantee the displayed loss. A stop-market trigger submits a market order when its trigger condition is met; the fill still depends on available liquidity. Spread, price gaps, latency, and a fast liquidation path can move the execution away from the trigger. The gap between stop and liquidation is context, not insurance.

A one-position calculator should not claim to reproduce cross margin. In isolated mode, the model has a defined margin amount assigned to one position. In cross mode, other positions, collateral, unrealized PnL, and account-level maintenance determine the boundary. Portfolio margin adds another layer of account-wide risk treatment.

Use an isolated estimate only when that matches the position you are modeling. If the live account uses cross or portfolio margin, inspect the venue's account-level output instead of forcing a single position into an isolated formula. HypeBasis deliberately withholds those modes from the standalone estimate.

A personal liquidation estimate and a market-wide liquidation map are different products. The first uses your visible scenario inputs. The second needs reliable position-level evidence. Open interest, leverage assumptions, recent liquidations, or price structure can suggest a risk zone, but they do not reveal every trader's hidden entry, margin mode, collateral, or stop.

If a heatmap derives clusters from assumptions, read it as a model. If it claims observed positions, verify the source, sample coverage, timestamp, and whether cross-account hedges are visible. A precise-looking band is not automatically precise evidence.

Start with position size, entry, and the amount of margin actually assigned to the isolated position. Calculate the tier-aware liquidation boundary. Add the planned stop and confirm which boundary comes first. Then inspect HYPE funding, current spread, visible depth, and recent movement; those do not change the maintenance equation, but they affect carry and the chance that an exit fills near its trigger.

Keep the inputs visible. A useful output should show liquidation price, distance from entry, liquidation notional, maintenance rate, tier, and deduction. If any of those are hidden, it is difficult to tell whether the tool used a flat shortcut or the current official table.

• Use the current HYPE margin table and record its source time.
• Model the margin mode you actually intend to use.
• Compare the planned stop with liquidation before calculating reward to risk.
• Treat stop fills, future funding, spread, and slippage as separate uncertainties.
• Recalculate after changing size, leverage, margin, or entry price.