Should the stop sit on the exchange or locally?

On the exchange. An exchange-side stop order still triggers if you lose connection, shut down or the platform is briefly unreachable; the platform places an exchange stop with every entry.

Price bounced right after stopping me out. What now?

That is normal and happens with any stop method. Do not immediately reverse into a chase or drop stops in future; judge a stop method on the statistics of dozens or hundreds of trades.

What stop-loss percentage should I use?

There is no fixed percentage. Find the structural stop, confirm it is wider than normal noise (for BTC, clearly wider than the average hourly range), and size the position from the stop distance.

Where to Place a Stop-Loss: Structural Stops, Minimum Distance and Trailing Take-Profit

How to set a stop-loss on crypto futures: why stops belong where the trade idea is invalidated rather than at a fixed percentage, why tight stops keep getting hit, when to activate a trailing take-profit, exchange-side stop orders and the four most common stop-loss mistakes. With real Bitcoin volatility data.

2026-10-26

Risk Management and Stop-Loss: Where to Place Stops, Risk-Reward and Avoiding Liquidation

A stop-loss belongs where price proves your idea wrong, not at a random percentage. Place the stop first, then size the position from the stop distance. Once in the trade, lock in gains with a trailing take-profit instead of moving the stop further away.

What a stop is for: capping the loss, not predicting

A stop is not a guess about whether price will turn back; it decides what you pay when the idea is wrong. Only with a stop on every trade is the loss per trade known, and only then can you size positions by risk and make long-run statistics meaningful.

So the order is: find the stop from market structure, then size the notional as acceptable loss ÷ stop distance. Not the other way round, deciding how big you want to be and then squeezing the stop to wherever you can afford the loss.

Structural stops: where the idea is invalidated

For a long, the stop goes below the most recent confirmed swing low; for a short, above the most recent confirmed swing high. If price breaks that low (or high), the structure behind the trade is gone and there is no reason to keep holding.

Unlike a fixed percentage, a structural stop adapts to the market: closer in calm trends, further away in volatile ones. The platform's signals use the structural level as the stop, and dynamic leverage then sizes the position from the stop distance so every stopped-out trade loses roughly the same amount.

Stops that are too tight get hit by normal noise

Across 2,160 one-hour candles of the OKX BTC perpetual from 2026-07-28 to 2026-10-25, Bitcoin's average hourly range (high minus low) was about 0.50%, and close to 0.9% around the US stock market open. A stop only a few tenths of a percent away can be taken out by an ordinary hour of trading.

That is why the platform requires an initial stop at least 1% away. In addition, for users with dynamic leverage the stop is randomly widened by 0–15% of its distance, so thousands of stops do not cluster at one price and get swept together; the position shrinks accordingly and the loss budget per trade stays the same.

After entry: trailing and fixed take-profit

The platform's recommended settings activate a trailing take-profit once the trade is 3.6% in profit. From then on it tracks the peak gain and exits after a 0.6% pullback from that peak, moving the exchange-side protective order up as it goes; if the move keeps running, a fixed take-profit closes it at 10%. Once active, it locks in roughly 3% at minimum.

There is also a filter before entry: if there is less than 3.5% of room to the nearest confirmed daily swing high (swing low for shorts), the trade is skipped. Every trade then has enough room to reach the trailing activation line instead of opening right under resistance.

The four most common stop-loss mistakes

One: moving the stop further away when the trade goes against you, turning a small controlled loss into a big one. Two: keeping the stop only in local software, which does nothing if your connection or computer goes down; stops belong on the exchange as conditional orders.

Three: placing the stop exactly on a round number or an obvious high or low, where stops cluster and wicks love to sweep; leave a little room beyond the structural level. Four: no stop at all, using liquidation as the stop. You do not choose the liquidation fill price, and you pay a liquidation fee on top.

#Stop-loss #Risk management

Should the stop sit on the exchange or locally?: On the exchange. An exchange-side stop order still triggers if you lose connection, shut down or the platform is briefly unreachable; the platform places an exchange stop with every entry.
Price bounced right after stopping me out. What now?: That is normal and happens with any stop method. Do not immediately reverse into a chase or drop stops in future; judge a stop method on the statistics of dozens or hundreds of trades.
What stop-loss percentage should I use?: There is no fixed percentage. Find the structural stop, confirm it is wider than normal noise (for BTC, clearly wider than the average hourly range), and size the position from the stop distance.