In equities you cannot trade overnight, so risk arrives as a gap and everyone faces the same constraint. In crypto the market runs continuously, so a position is live and manageable at every hour — by anyone who happens to be awake.
You are not. Which means, in practice, most positions spend most of their life unmanaged.
The asymmetry
Your stop is not watching in the way you would be. It executes at whatever the book offers, and the hours you are asleep are disproportionately the thin ones where spreads are widest and a move travels furthest.
Meanwhile the liquidation engine runs at full attention, judged on the mark price, indifferent to your sleep schedule.
Sizing from the unattended case
The useful question is not how much you can afford to lose, but what happens to this position in the worst eight hours you are not looking at it. Size so that answer is survivable, then the attended hours take care of themselves.
Concretely that means leverage low enough that an ordinary bad night does not approach liquidation, and it means treating weekends as their own case rather than as more of the same.
Why hard stops are not the whole answer
A stop bounds the loss and executes into whatever exists. In the thin conditions where you most need it, that fill can be materially worse than the level — the round-trip asymmetry in the true cost of a round trip.
So a stop is a limit on how wrong things get, not a guarantee of the price. Sizing has to assume the fill is worse than the level, particularly overnight.
The one that scales with everything else
Be smaller when system leverage is high, because that is when an ordinary move becomes a cascade. Sizing is the only defence that works in every one of these scenarios at once, which is why it is the thing to get right rather than the thing to optimise last.