The trade-price chart never reached the stop, yet the position entered liquidation. Those observations can be compatible when they refer to different prices.
Crypto mark price is a venue-defined risk reference used for purposes such as unrealized profit and margin calculations. It is distinct from the latest trade and from an executable bid or ask. To assess liquidation risk, identify the product's margin mode and authoritative risk measure, then compare the stop's actual trigger source with that mechanism.
This AI-assisted article uses official Bybit examples of documented mechanisms and an original hypothetical price timeline. It does not estimate an actual account's liquidation threshold or recommend leverage, collateral levels or a particular trigger setting.
Three numbers with different jobs
The last traded price records a completed transaction in the market. An index price represents a defined reference constructed from specified underlying price inputs. A mark price applies the venue's methodology for the particular derivative product. None of those names, by itself, promises a quantity can be bought or sold there now.
Bybit describes mark price as a reference for unrealized profit and liquidation and publishes different methodologies and fallback conditions. A formula copied without its product scope can become misleading as rules change. Bybit: mark price for perpetual and expiry contracts.
An actual exit still interacts with available liquidity and the order's terms. A market sell may consume bids at several prices; a limit may not fill. Marking a position at a reference price is an accounting observation, not an execution receipt.
For this reason, a dashboard should name the price source next to a threshold. “Price is safely above liquidation” is incomplete if it silently compares a last trade with a mark-based isolated-position threshold.
A stop can be untouched while the risk threshold is crossed
Construct an isolated-margin, USDT-linear long equivalent to 0.020 BTC. Its entry is 64,000 USDT/BTC. Assume the exchange's authoritative liquidation price for this example is 62,600 and remains unchanged during the two observations. This number is an input to the illustration, not a liquidation formula derived from leverage.
The operator's sell stop activates on last traded price at 62,700. The following observations are invented and deliberately show the two sources diverging:
| Observation | Last traded price | Mark price | What the two conditions show |
|---|---|---|---|
| Earlier | 63,200 | 63,000 | Last-price stop not reached; mark above the stated threshold |
| Later | 62,720 | 62,580 | Last-price stop still not reached; mark below the stated threshold |
At the later observation, 62,720 > 62,700, so the selected last-price stop condition has not been met in this constructed timeline. But 62,580 < 62,600, so the mark has crossed the example's isolated-margin liquidation threshold.
Bybit documents the possibility of a last-traded-price stop being preceded by mark-price liquidation. The event sequence must be evaluated against the right source, rather than treating a candle from one series as evidence about the other. Bybit: order execution and liquidation FAQ.
The example does not promise that selecting mark as the stop source would avoid liquidation. A stop must still activate and its resulting instruction must execute. A price gap, unavailable liquidity, account constraint or processing delay can prevent the intended result. Our stop-market versus stop-limit guide separates those later stages.
Distance is a measurement, not a countdown
At the earlier mark of 63,000, the distance above the assumed 62,600 liquidation price is 400 USDT/BTC. Relative to that mark, it is 400 ÷ 63,000 × 100 = approximately 0.6349%.
Using the last trade of 63,200 would produce a 600 USDT/BTC gap instead. That is a larger number because it is a different observation, not because the account gained additional collateral. Record the source and denominator with any distance metric.
For a long, the simple positive difference is mark minus an applicable liquidation price. For a short, the directional comparison reverses. If the account's risk is governed by an aggregate margin measure rather than a fixed position price, that simple distance may not represent the actual trigger at all.
A percentage gap also says nothing about how long it will take to close. A small move can occur abruptly; a large move can occur through a discontinuous book. Do not convert the distance into a promised response time without a justified model and its limitations.
In the constructed long, marking 0.020 BTC from a 64,000 entry to 62,580 gives 0.020 × (62,580 − 64,000) = −28.40 USDT of gross unrealized price change. That arithmetic is not an estimate of the final liquidation charge or the account's entire loss. Fees, funding, collateral rules and actual liquidation handling are absent from it.
Margin mode changes what the threshold means
A per-position liquidation price can be useful in the correct isolated-margin context. Cross and portfolio arrangements require a broader view of account equity, maintenance requirements and other exposure. One visible position cannot necessarily explain an account-wide risk event.
Bybit's FAQ distinguishes isolated mark-price triggering from cross and portfolio maintenance-margin-ratio triggering. Its position API also documents that a liquidation-price field can be estimated or unavailable depending on mode and conditions. A blank field is not proof that liquidation cannot occur. Bybit API: position information.
This is why multiplying entry by one minus reciprocal leverage is insufficient as a general liquidation calculator. It can omit maintenance requirements, fees, collateral effects, other positions and product-specific rules. A rough arithmetic illustration should never be presented as the authoritative exchange trigger.
Account resources can change while the market is quiet. Review actual funding payments, charges and other activity alongside the relevant risk state. The perpetual funding-cost ledger separates those cash flows from a price-only profit calculation.
A displayed threshold is also an observation at a particular time. If collateral or positions changed after it was captured, reusing the old screenshot can misdescribe the state that later reached the engine.
Diagnose a disputed event with matching evidence
Begin with the exact exchange, account, contract, margin mode and position side. Record the event time and timezone. Then retrieve the actual order configuration, including its trigger source, rather than infer the setting from the chart currently open.
Collect the relevant mark history, last-trade history, order status and actual fills. Keep each series identified. Bybit provides a separate mark-price candlestick endpoint; requesting an ordinary trade-price candle is not the same data query. Bybit API: mark-price candles.
A candle is still a summary over an interval. It may show that a threshold was within a range without establishing the exact order of every trigger, fill and account update. For a close dispute, use supported event records where available and keep any missing sequence explicit.
Candle-based backtests face the same limitation about intrabar ordering. A chart that combines high, low, open and close cannot by itself prove which event occurred first for an individual account.
Reconcile any partial reductions and their timestamps too. The partial-fill guide explains how an incompletely closed position can remain exposed even when a local exit attempt looks complete. A requested close quantity is not the same as the quantity executed before the risk event.
A mark trigger still needs an execution review
Choosing the same reference for a stop as the venue uses for an isolated-risk comparison can remove one source mismatch. It cannot reserve liquidity or make the trigger price a guaranteed fill.
If the resulting instruction is a limit, inspect which bids or asks are eligible after activation. If it is a market instruction, inspect the actual fills and any venue price protections or cancellations. The order-book-depth worksheet illustrates why one reference price cannot describe the cost of a larger close.
Preserve the stop's identity and final outcome after any manual intervention. Another process may have changed the position, canceled an exit or submitted a competing reduction. An incident record should explain those actions without claiming that an attempted cancel or reduction definitely succeeded.
A workflow pause also does not establish that the exchange position is closed. The bot kill-switch guide treats stopping new work, handling existing orders and verifying exposure as distinct jobs.
Keep the risk view and the trading view connected
For an operational review, put the instrument, price source, observation time, margin mode, position side and relevant risk metric together. Show the stop's source and resulting order type alongside them. This makes a mismatch visible before an operator relies on a reassuring chart.
For a historical review, distinguish actual source records from constructed explanations like the timeline above. If mark history or account state is missing, state the limit. Do not fill that gap with an assumed exchange defect or a guarantee that a different setting would have saved the position.
KLINEO's trading risk disclosure covers wider execution and service limitations. The role of mark-price analysis is to explain which observation governs which condition. It cannot make leverage safe, ensure time to react or guarantee that an order executes before liquidation.