Bybit Auto-Deleveraging: Insurance Funds and Queue Ranking

What ADL is for

A perpetual futures position has to absorb losses when a leveraged position fails. The first line is the liquidation engine, which closes the position. The second line is the insurance fund, which covers any shortfall between the liquidation proceeds and what the position owed. The third line is auto-deleveraging: if the fund itself is under stress, the venue closes opposing positions to restore the fund. The difference between the liquidation price and the bankruptcy price — the two thresholds that decide how much the fund must absorb — is covered in the liquidation price versus bankruptcy price guide.

Bybit describes ADL as a system that “helps control the platform’s overall risk during liquidation events in extreme market conditions where the insurance fund cannot cover excessive liquidation losses”, working “by automatically deleveraging profitable or highly leveraged positions on the opposite side of the liquidated positions, based on the ADL ranking.”

Three words carry the design: automatically, opposite side, ranking. The venue does not ask; it selects counterparties by a published score and closes them.

Trigger one: the 8-hour drawdown

Bybit’s first trigger is a drawdown test on the insurance fund itself.

• Trigger: ADL fires when the drawdown of a single trading pair over the past 8 hours “reaches or exceeds the 8H PnL Drawdown Trigger Line of the insurance fund’s highest balance” in that period.
• Execution: the venue calculates a bankruptcy price from the trigger line and settles affected positions there.
• Stop: ADL stops when the drawdown ratio falls to or below the 8H PnL Drawdown Stop.

Bybit’s own example makes the arithmetic concrete. An ABCUSDC insurance fund peaked at 20,000 USDC in the past eight hours, with a 30% trigger and a 25% stop. Open interest carries 1,000 USDC of position margin and 8,000 USDC of realised and unrealised losses. The drawdown ratio is:

(8,000 − 1,000) ÷ 20,000 = 35%

That is above the 30% trigger, so ADL fires. The fund treats 30% as the permitted drawdown, which is 6,000 USDC, and the excess loss is 8,000 − 6,000 = 2,000 USDC. Users compensate that amount, bringing the ratio back to 30%. ADL then continues until the ratio falls to 25% or below.

The formula shows why the trigger is not simply “the fund is empty”. A fund can hold a positive balance and still be in ADL, because the test is a drawdown from its recent peak, not a level.

Trigger two: negative combined balance

The second trigger covers pairs with independent pools.

• Trigger: the combined balance of multiple pairs with independent insurance funds falls to or below zero.
• Execution: the venue treats the combined balance as zero, redistributes the fund across pairs by each pair’s profit and loss ratio, and recalculates the bankruptcy price.
• Stop: the combined balance rises above zero.

The redistribution is the subtle part. A pair with a positive pool can be deleveraged so its surplus can be moved to pairs with negative pools. The venue then re-checks both conditions and repeats until the stop condition is met.

The queue: leveraged return

The ranking is the part traders can influence, so it is worth reading precisely.

Bybit’s rule: “The ADL ranking is based on the leveraged return. Accounts with higher leveraged returns are ranked higher.” Opposing positions at a loss may still be selected, but profitable ones are given priority.

The formulas differ by margin mode.

Isolated margin

profitable position: leveraged return = position PnL% × position margin rate
losing position:     leveraged return = position PnL ÷ position margin rate

where position PnL% is (mark − entry) ÷ entry for a long and (entry − mark) ÷ entry for a short, and the position margin rate is maintenance margin ÷ (initial maintenance margin + additional margin + unrealised PnL).

Cross or portfolio margin

profitable position: leveraged return = position PnL% × account MMR
losing position:     leveraged return = position PnL% ÷ account MMR

using the account maintenance margin rate rather than the position’s.

Two consequences follow from the algebra.

1. Leverage dominates. A small profit on a thinly margined position can outrank a large profit on a heavily margined one, because the margin rate multiplies the percentage return. This is why “lower your leverage” is the first line of Bybit’s own advice for reducing ADL exposure.
2. The ranking is dynamic. A position’s mark price, margin and PnL all move, so the score changes continuously. The five-light indicator on the position is a snapshot, not a fixed property of the trade.

The published ranking example

Bybit’s article uses six short positions and ranks them by leveraged return:

If the insurance fund takes over 5,000 long contracts, Trader A is selected first and 5,000 of A’s 5,500 contracts are closed. A keeps 500 contracts and, after the reduction, may no longer rank first. If 10,000 contracts need to be deleveraged, A, B and C are all selected.

The example shows why position size is not the ranking. F holds 5,000 contracts — nearly as many as A — and ranks last. A holds the most and ranks first, because the ranking measures return per unit of margin, not size.

What a trader actually receives

Three details determine the economics of being selected.

• Price. The fill is at the bankrupt position’s bankruptcy price. Bybit states that any difference between that price and the market price contributes to the insurance fund. A trader closed at the bankruptcy price may therefore receive less (or more) than the last traded price.
• Fee. “A maker fee will be charged to the ADL traders whose positions are reduced, while a taker fee will be charged to the trader whose liquidation triggered the ADL.” The deleveraged trader pays the maker rate even though the closure was involuntary.
• Operational consequences. All active orders are cancelled, and the trader is notified by email and/or SMS. The position can be re-entered immediately, but the orders that were working around it are gone.

That combination — maker fee, bankruptcy-price fill, cancelled orders — is the reason ADL is a risk to model rather than a rare event to ignore. The maker charge is the ordinary maker rate applied to an involuntary close; ADL does not introduce a special fee tier, it changes who pays which side of the normal one.

How to reduce exposure, honestly

Bybit’s own list is short:

1. Lower leverage. This lowers the ADL ranking in real time.
2. Partially close a profitable position. This reduces the number of contracts exposed but “won’t lower your ADL ranking”.
3. Fully close the position if partial closing is not enough.

The asymmetry in step 2 is the useful insight. Reducing size does not change the score, so it reduces the amount at risk per event but not the probability of being selected. Only leverage changes the ranking.

Where the fund data lives

Bybit publishes insurance pool balances through the API and the Insurance History page. The help article states that isolated pools update every minute and shared pools every 24 hours, and that WebSocket pushes for isolated pools arrive every second while shared pools have no push updates.

A trader monitoring fund stress should therefore not treat a 24-hour-old shared-pool figure as current. The update cadence is part of the data’s meaning.

Funding and ADL are separate mechanisms that feed the same margin balance: a funding payment settles into the position’s PnL, which is one of the inputs to the ADL ranking. The funding fee worked example shows how that accrual is calculated.

Risk note. Trigger conditions, ranking formulas, fill price, fees and the worked example were read from Bybit’s ADL help article on 2026-10-28. Trigger lines and stop levels are pair-specific and published on Bybit’s insurance fund page; they can change. This article describes the mechanism and is not trading advice.