Auto-deleveraging: when a winning position closes itself

The position was in profit. There was no stop nearby, the liquidation price was far away, and nothing on the chart had turned. Then a notification: position closed, all open orders cancelled, filled at a price that never appeared on the candle you were watching. The channel posts "target hit" an hour later, because from where the channel is standing, it was.

What closed the position is called auto-deleveraging. It is not a bug, not a stop-hunt, and not the exchange taking your profit. It is the last mechanism in the chain that keeps a leveraged venue solvent, and it is the one part of a follower's outcome that no discipline on your side can prevent.

Why would an exchange close a position that is winning?

Start one step earlier, at the losing side of the trade. When a leveraged position runs out of margin, the venue closes it — and it does not close it at the market price. Bybit's help centre states that when a position is liquidated "it will always be settled at the bankruptcy price, which is the price level where there's no initial margin left, regardless of the current market price" (Bybit, Insurance Fund). If the liquidated position can actually be sold into the book at a better price than that, the leftover margin goes into the insurance fund. If it cannot — thin book, violent move, everyone offside at once — the fund pays the difference. That is what the fund is for. The mechanics of the losing side are covered separately in liquidation explained; this article only picks up where that fund runs out.

Because it can run out. Bybit describes the case plainly: if the insurance fund is insufficient to cover the excess losses, "they will be absorbed by the ADL system, which means the profits from other traders' positions on the platform will be used to cover the shortfall."

Binance frames the same step the same way: "Auto-Deleveraging (ADL) is the final step in the liquidation process and occurs only if the Futures Insurance Funds are unable to accept a bankrupt futures position" (Binance, What Is Auto-Deleveraging (ADL) and How Does It Work). OKX calls it "the final liquidation process deployed to protect the OKX security funds when a fund isn't able to further takeover additional bankrupted positions" (OKX, Introduction to Auto-deleveraging).

The logic is arithmetic, not malice. A derivatives venue is a closed loop: every unit of loss on one side is a unit of profit on the other. If a bankrupt position cannot be closed for what it owes, the shortfall has to land somewhere. The venue's own reserve absorbs it first. When that is not enough, the only remaining pool of money is the profit sitting in open positions on the opposite side — so the venue closes some of them and offsets the books.

Who gets taken first?

Not at random, and every major venue publishes the rule.

Bybit's ADL mechanism page says the system "works by automatically deleveraging profitable or highly leveraged positions on the opposite side of the liquidated positions, based on the ADL ranking," and defines that ranking this way: "The ADL ranking is determined based on the leveraged return of a position. Please note that opposing positions that are at a loss may still be selected, but those in profit are given priority." Its published formula for a position in profit in isolated margin is the position's PnL percentage multiplied by the position margin rate.

Binance: "Positions that are more profitable and more highly leveraged will be queued for ADL liquidation first, ahead of positions that are less profitable and less leveraged." Its published ranking, for a position in profit, is PNL Percentage multiplied by Effective Leverage, where PNL Percentage is unrealised profit over the absolute position notional, and Effective Leverage is the absolute position notional over account balance plus unrealised profit.

OKX: positions in the queue "are ranked based on the leverage PnL%," which for profitable positions it defines as the position's unrealised PnL percentage divided by the account-level maintenance margin ratio.

The three formulas are not identical and the terms are not interchangeable between venues. What they agree on is the shape: profit multiplied by leverage. Two things put you at the front of the queue, and they are the two things a leveraged winner is made of.

All three also show you where you stand. Binance states that "the interface for each futures position will include an indicator of the likelihood that the position will be subject to ADL liquidation." OKX describes a light with one to five bars, where "more filled bars indicate your position being closer to the front of the ADL queue, if there is an ADL event." Bybit tells traders to "refer to the ADL priority lights on their positions to estimate their ADL rankings." That indicator sits on the position row, and most people who follow signals have never looked at it. It is worth knowing which corner of your screen it lives in before the day you need it.

What price do you actually get?

This is the part that explains the number in your account, and it differs by venue.

On Binance, "the profitable futures position will be closed at the Bankruptcy Price of the liquidated order for the losing trader subject to liquidation" — the price at which the other trader's margin balance reached zero, not the price you were watching. The same page warns that "the Bankruptcy Price may be out of the contract's market price range."

Bybit is the same in effect: "Selected positions will be settled at the bankruptcy price of the liquidated positions taken over by the system."

OKX splits the case: "In normal circumstances, deleveraged positions are forcibly closed at the mark price at the time of matching. However, if the applicable security fund is nearly depleted, the deleveraged positions could be closed at bankruptcy price."

Fees differ too, in both directions. Binance: "No trading fee will be charged to the trader whose profitable position is subject to ADL." OKX: "Trading fees are not charged on deleveraged positions, while liquidation fees apply to the liquidated orders." Bybit publishes the opposite treatment: "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."

Three venues, three rulebooks, all current as of the pages checked in August 2026 — and all of them subject to change without your inbox being consulted. That is the general lesson about venue mechanics, and it applies well beyond ADL: the rules that decide your result are the rules of the exchange you are on, which is the wider point in what exchange risk actually means for a follower. Never carry one venue's number over to another.

Why does this land on signal followers specifically?

Because ADL fires exactly where a channel's calls tend to cluster.

Auto-deleveraging is triggered by a liquidation cascade the book could not absorb. That requires a violent move against a crowded side, usually in something with thin depth. A signal channel produces, by design, a crowd of accounts entering the same coin in the same direction within minutes of each other, frequently in a small-cap pair, frequently at leverage the post did not specify. If the move goes the channel's way and the other side blows up en masse, the followers who were right are sitting in exactly the bucket the queue sorts to the top: profitable and leveraged.

And leverage is doing double duty here. It is the multiplier in every published ranking formula above, and it is also the thing that made the move big enough to break the book in the first place. What leverage does to a losing streak covers the arithmetic on the losing side; this is the same dial working against you on a day you were right.

One venue-specific note worth carrying: Binance states that "coin-margined contracts are more likely to be subject to ADL than USD-margined contracts," because coin-margined contracts sharing the same collateral asset share one insurance fund. If a channel posts coin-margined pairs, that is a fact about the instrument, not about the call.

Why will the channel's numbers and yours disagree?

Because both records are true and they are recording different events.

The channel's scoreboard tracks the price. Price reached the target, so the trade goes down as a win at the published percentage. Your account tracks the fill. Your position was closed early, at someone else's bankruptcy price, before the target printed. Neither party lied. The gap is a mechanic, and it is invisible in any results post because a results post has no field for it.

This is the same class of divergence as taking TP1 and then getting stopped out, where the channel books the target and the follower books a scratch. What makes ADL harder is that there was no decision on your side to review afterwards. You did not size wrong, exit early, or panic. There is nothing to correct.

The practical consequence: when you evaluate a channel, judge it on what the calls did, and keep your own ledger of what your account did, separately. Treating your fills as evidence the channel is faking is as wrong as treating the channel's price-based percentage as a forecast of your balance. Two honest records, two different questions.

What actually reduces exposure?

The honest list is short, and it comes from the venues rather than from anyone's strategy.

Bybit: "You can reduce your exposure to ADL by lowering your leverage or partially closing your positions in profit," adding that "lowering the leverage will lower the ADL ranking in real time," while "partially closing a highly leveraged position can reduce the number of contracts exposed to ADL risk, although it won't lower your ADL ranking."

OKX publishes four: add margin to lower account effective leverage; reduce positions to lower the maintenance margin requirement; prioritise major coins and contracts with deeper liquidity; and use its delta-neutral strategy, since for futures positions that are fully or partially hedged "the hedged portion of the position will be ranked lower in the ADL queue."

Notice what is not on either list. There is no order type that opts out, no stop placement that helps, no timing trick. Watching the indicator does not prevent anything — it only tells you where you stand. OKX is explicit that beyond its API channel, "other than the above, no warnings are issued prior to ADL occurring." Reduce leverage, reduce size, prefer deeper books. That is the whole set.

If it happens to you

Three practical things, all published by the venues.

Your open orders are gone. Binance cancels "all open orders for all tokens (in cross margin mode) or for the same token (in isolated margin mode)." Bybit cancels all active orders and notifies by email. If you had a stop and a target resting for a position that no longer exists, they went with it — and if you are running several positions in cross margin, check the ones you were not thinking about.

You are flat, not paused. OKX puts it directly: "Deleveraged positions won't gain any further profit without you placing a new order into the order book. A key risk is that any new entry price for new orders may be different from the deleveraged positions'." Re-entering is a fresh decision at a fresh price, into a market that just moved violently enough to break someone.

Log it as its own category. Not a stop, not a liquidation, not a discretionary exit — an ADL fill, with the venue, the pair, and the fill price. Otherwise, six months on, it reads in your own records as a trade you closed early for no reason, and you will draw the wrong conclusion about yourself instead of the right one about the instrument.

None of this is a reason to distrust the exchange that closed the position. A venue with no such mechanism is not safer; it is a venue where the shortfall has nowhere to go. It is a reason to know that a follower's result and a channel's result can differ for reasons that have nothing to do with the quality of the call — and that the only dial you hold here is the same one you already hold everywhere else: how much leverage is riding on being right.