Separate position margin from account mode
Cross versus isolated describes how a position uses collateral. Account abstraction mode describes how spot, perps and different DEX balances are organized. Confusing the two can produce incorrect assumptions about which losses or balances interact.
The official account modes are unified, portfolio margin and manual or standard. A legacy DEX-abstraction mode is being discontinued and is not recommended for new integrations.
Cross, isolated, strict isolated and no-cross
Cross margin shares eligible collateral with other cross positions, improving capital efficiency while allowing one loss to consume a common buffer. Isolated margin confines assigned collateral and permits margin adjustments after opening.
Strict isolated prevents manual margin removal, with margin released proportionally as the position closes. HIP-3 can also expose no-cross behavior, which keeps isolated accounting while allowing margin removal but disallows cross margin.
| Mode | Collateral scope | Primary trade-off |
|---|---|---|
| Cross | Compatible cross positions | Efficient but interconnected |
| Isolated | One position | Contained but less flexible |
| Strict isolated | One position; no manual removal | Tighter containment |
| HIP-3 no-cross | Isolated with removable margin | No cross netting |
Initial and maintenance margin
Initial margin for a position is position size multiplied by mark price and divided by selected leverage. Maximum leverage is asset-specific. Maintenance margin is currently half the initial margin requirement at maximum leverage under the documented standard schedule.
Removing collateral must still satisfy initial-margin rules and a transfer requirement based on at least 10% of total open notional. Mark price and account equity—not a single last trade—drive liquidation logic.
initial_margin = position_size × mark_price ÷ leverage
Choose by failure containment
Cross margin is not automatically safer because it has more collateral, and isolated margin is not automatically safer because it limits spillover. The correct choice depends on correlated positions, operational monitoring and what should happen when one thesis fails.
Keep a buffer above maintenance, place risk exits before the liquidation boundary and verify the active mode after every account-setting change.