At 20x the naive expectation is 5% of room. The real distance is shorter, it shrinks while you hold, and in the signal posts we parsed the stop often sits past it - so the position dies before the stop is ever reached.

What has to happen before you are liquidated?

Your equity has to fall below the maintenance margin: the minimum the venue requires you to keep behind an open position. When it does, the position is closed for you, at a price you did not choose, and the margin behind it is gone.

That is the whole trigger. It is not a penalty and it is not discretionary. The venue is protecting itself from holding a position whose losses have grown larger than the money backing it, and the moment your equity crosses that line, the close happens automatically.

Bybit's help centre states the trigger as the maintenance margin rate reaching 100%. OKX words it as the maintenance margin ratio reaching 100% or below, at which point position reduction or forced liquidation may be triggered. Two venues, two phrasings, one mechanism.

What follows is where that line actually sits, why it sits closer than most followers assume, and which parts of the distance you decide yourself.

Why does 20x leave less than 5% of room?

Because the maintenance margin is subtracted from your room before you get any of it. The leverage number tells you how much of the position you funded. It does not tell you how much of that funding you are allowed to lose.

The arithmetic is short enough to do on this page. For an isolated position, with no margin added later:

adverse move to liquidation = 1 / leverage - maintenance margin rate

The first term is the room the leverage figure implies. The second is the slice the venue keeps. Using 0.5% as an illustrative maintenance margin rate, which is roughly where the lowest risk tier sits on major venues as of August 2026:

Leverage What the number implies Where liquidation actually sits
5x 20.00% 19.50%
10x 10.00% 9.50%
20x 5.00% 4.50%
25x 4.00% 3.50%
50x 2.00% 1.50%
100x 1.00% 0.50%

Read the bottom rows carefully. The subtraction is a fixed half a percent, so it barely dents the 5x row and takes half the room off the 100x row. The higher the leverage, the larger the share of your buffer that was never yours.

One number in that table is illustrative and has to be treated that way: the maintenance margin rate is tiered, and it rises with position size. Bigger position, worse tier, shorter distance at the same leverage.

What eats the distance while you wait?

Fees and funding, both deducted from the same margin that is holding the position open. The distance in the table above is the distance at the moment you open. It shrinks every hour you hold.

Three separate drains, and none of them require the price to move against you:

  1. The opening fee comes out of margin immediately.
  2. Funding is exchanged every few hours for as long as the position is open, and on a crowded side it runs against you consistently. What that costs over a multi-day hold is the whole subject of Funding rates explained.
  3. The tier, if you add to the position. Size up and the maintenance margin rate steps up with you.

The practical consequence is that a position which opened with 4.5% of room does not still have 4.5% of room on day three. A signal that sits waiting for its entry to work out is paying for the privilege out of the buffer that keeps it alive.

Which price triggers it, the one on your chart?

No. Liquidation is checked against the mark price, an index-anchored fair price built from several references, while the chart in front of you is usually showing the last traded price on that venue.

The reason for the split is sound, and it protects you more often than it hurts you. Suppose liquidation ran on the last traded price. One large order on a thin book could then wick the price down for a second and liquidate every position inside the wick, setting off a cascade from an event no other market ever saw.

The cost of that design is a gap you cannot see. Two prices, one on your screen and one running the trigger, and they disagree exactly when the market is disorderly, which is exactly when your position is closest to the line. "The wick did not really happen" can be true of your chart and false of the number that closed you.

Cascades are the large-scale version of this, and the October 2025 event is covered in Spot, futures and perpetuals explained.

What leverage do signal posts actually ask for?

Enough to put liquidation inside the range of an ordinary day's movement. In the channel archives held in this repository, 385 posts state a leverage figure explicitly, and the median is 25x, as of August 2026.

The distribution, from work/liquidation-explained/parse_leverage.py, which is in the repository so the parse can be checked rather than trusted:

Leverage asked for Posts Share Liquidation distance
5x 61 15.8% 19.50%
10x 45 11.7% 9.50%
20x 71 18.4% 4.50%
25x 108 28.1% 3.50%
50x 83 21.6% 1.50%
Other levels 17 4.4% -

Just under seven in ten of these posts ask for 20x or more. Around a fifth ask for 50x or more, where a 1.5% move against the position ends it.

Two limits on that figure, both of which matter. The 385 posts come from 12 channels, and four of them supply 351 of the 385, so this describes what those feeds ask for and not a market-wide average. And 62 posts state a range rather than a level ("5x_10x"); those are counted at the low end, which makes the distribution above the conservative version of itself.

Does a channel's own stop survive its own leverage?

Often it does not. When the stop is further from the entry than the liquidation point, the position is closed by the venue before the stop is ever reached, and the stop published in the post is decoration.

In the subset of posts that state a leverage, an entry and a stop together, 101 could be measured. The median stop sits 5.04% from entry and the median leverage is 20x, which puts liquidation at roughly 4.5%. Those two medians are on the wrong side of each other.

How many individual posts land that way depends on how you read them, and the honest answer is a range rather than a number:

Assumed fill Share whose stop sits beyond liquidation
Far edge of the entry zone 14.9% to 30.7%
Average of the entry zone 35.6% to 53.5%
Near edge of the entry zone 54.5% to 63.4%

Across all 18 variants of the parse the script runs, the share moves between 14.9% and 63.4%. A spread that wide disqualifies any single figure from being quoted as a measurement, so none is quoted here. Three things do survive every variant:

  • The share is never small. Even the most generous reading leaves one in seven of these posts with an unreachable stop.
  • Where you fill inside the entry zone decides it. The same post is survivable filled at one edge and hopeless filled at the other, which is a reason to care about entry quality covered in Order types for signal followers.
  • Fees and funding are excluded from all of it, and they only ever move liquidation closer. Every share in that table is a floor.

What does liquidation cost beyond the margin?

More than the margin, which is the part that surprises people. The forced close carries a fee, and it is charged at the taker rate because a liquidation is a taker order on your side.

Two prices matter once the process starts. The liquidation price is where your equity stops being sufficient. The bankruptcy price is where it reaches zero. The venue tries to close your position in the gap between them, and what happens to that gap is worth knowing:

  • Closed better than the bankruptcy price, and the remainder goes to the venue's insurance fund.
  • Closed worse, and the insurance fund covers the shortfall.
  • If the fund cannot cover it, auto-deleveraging closes profitable positions on the other side to balance the books, starting with the most profitable and most leveraged. This is a last step and it is rare, but it means a correct position can be closed for you because of somebody else's loss.

Some venues reduce a position in stages instead of closing it at once, continuing to trim if the remaining position still fails the requirement for its new tier. That is better than a single forced close, and it is not universal, so it is not something to count on at the moment you need it.

Why does the position not come back when the price does?

Because liquidation is a completed sale, not a suspension. The recovery you watch afterwards happens to a position you no longer hold.

This is the most expensive misunderstanding in the whole subject. A stop and a liquidation both close a trade, so they feel like the same category of event, and they are not. A stop is a decision you made in advance, at a distance you chose, with the rest of your account intact. A liquidation is the venue closing you at whatever the book offers, having first taken the maintenance margin and the fee.

The asymmetry compounds over a losing run, because a liquidated position removes the capital that the next trade was going to be sized against. What leverage does to a losing streak works through what a normal run of losses does to an account, and what account blow-ups really mean covers the end state.

What actually moves liquidation further away?

Position size, not leverage. This is the part that gets inverted most often, and getting it the right way round changes how a follower reads every signal.

Leverage sets the distance to liquidation as a percentage of the entry price. Size sets how much money is behind the position in the first place. A follower who halves the size and keeps the leverage has the same liquidation distance and half the exposure. A follower who halves the leverage and doubles the size to compensate has changed nothing about the risk and moved the liquidation point.

Which gives an order of operations that survives whatever a signal says:

  1. Decide the loss you accept on this trade, in money, before reading the entry.
  2. Work back to a size from that number and the distance to the stop. The arithmetic is in Position sizing for signal followers.
  3. Check the leverage the post asks for against the stop it published. If liquidation sits nearer than the stop, the post's own plan cannot execute, and that is a fact about the post rather than about the market.
  4. Treat a wide entry zone as a range of outcomes, because where you fill decides how much room the trade ever had.

Step three is the one that costs nothing and is skipped most. It is a single subtraction, it uses only numbers the post already gave you, and it tells you whether the trade as written can do what it claims. When to skip a signal covers the related case where a call has simply moved too far to take.

What this does not prove

The channel archives here are a sample, not a census, and the leverage figures come from 12 channels with four of them dominating the count. Nothing above establishes what a typical signal channel asks for across the market.

The maintenance margin rate used throughout is illustrative. Real rates are tiered, differ by venue and by contract, and rise with position size, so every distance quoted above is an approximation whose direction is reliable and whose second decimal place is not.

The verification behind the mechanical claims is weaker than usual, and that belongs on the page rather than in a working note. This run could reach search results but could not open the venue help pages themselves. The mechanics above therefore rest on consistent descriptions from several independent publishers, not on a primary page read directly. Where a claim could not clear that bar, it was left out.

And none of this says anything about whether a given call was any good. A well-reasoned trade at 50x and a poor one at 50x have the same liquidation distance.

The practical read

Liquidation sits closer than the leverage number implies, moves closer while you hold, and triggers on a price your chart is not showing you. All three are properties of the instrument, and none of them depend on the signal being wrong.

The two decisions that matter are both yours and both made before the trade. The first is size, because it sets what a liquidation actually costs you. The second is the subtraction in section two, because it tells you whether the post's own stop is reachable at the leverage the post is asking for. As of August 2026, in the posts we could measure, that check fails often enough that it deserves to be a habit.

Whether a channel's calls are worth following at all is a separate question with its own evidence, and the current ratings for the channels we can score are at Signal Providers. Nothing here recommends any asset, any venue or any level of leverage.

Sources