To a first approximation, the adverse move that wipes your posted margin is the inverse of the leverage: 10x means roughly 10%, 20x roughly 5%, 50x roughly 2%. In practice it is tighter, because you are liquidated at maintenance rather than at zero, and fees come out too.
The curve is the point
Read those numbers as a sequence rather than individually. Going from 2x to 4x costs you 25 percentage points of room. Going from 20x to 40x costs about 2.5. In room terms the second change is trivial; in the way people talk about leverage it sounds like the same doubling.
So the interesting region is entirely at the low end. Above roughly 20x the position is already living inside ordinary intraday noise, and further increases barely change a survival profile that is already governed by luck.
What makes it tighter than the arithmetic
Three things, all of which move the wrong way at once.
Maintenance margin means you are closed before the equity is gone, and the requirement rises with position size. Funding is deducted while you hold, so a position on the crowded side is paying rent that erodes the buffer without price moving at all. And it is all measured against the mark price, not the chart.
The right way to size
Backwards, from the level rather than from the multiple. Decide where price would prove the idea wrong, put the liquidation meaningfully beyond it, and let that determine the leverage — instead of picking a multiple and discovering afterwards where the liquidation landed.
This matters more in a thin market than a deep one, because the move that reaches you is smaller and, per thin liquidity and volatility, more likely to extend once it starts.