Place your stop loss where your trade idea is proven wrong, add a buffer for normal price noise, and only then work out the position size so that hitting the stop costs a small, fixed share of your account. Most traders do it the other way round: they pick a size first and set the stop at whatever loss feels bearable. That puts the stop in the wrong place, usually too close.
How close is too close? We measured it. Over the past year of 4-hour candles, a stop 2% away was touched within 24 hours on 29% of test entries on Bitcoin, and on 47% on Solana and Dogecoin, with no trade idea behind the entries at all. That was just ordinary movement.
- The trade idea decides the stop, and the stop decides the position size. Never the other way round.
- A fixed 2% stop was touched within 24 hours on 29% of BTC test entries and 47% of SOL and DOGE entries, from Nov 2025 to Nov 2026.
- Measured in ATR, the six coins behaved almost the same: a 1 ATR stop was touched 47 to 49% of the time, a 2 ATR stop 20 to 22%.
- Stops tend to cluster just beyond obvious levels and round numbers, so leave a buffer past them.
- Never widen a stop after entry. Tightening it is fine when the rule was set before the trade.
The order matters: idea, stop, size
A stop loss has one job: to get you out when the reason you entered no longer holds. That gives a simple order of work for every trade:
- Find the invalidation point. Where would the chart prove your idea wrong? For a long from support, it is a clear break below that support. For a breakout, it is price falling back inside the pattern.
- Add a buffer. Price wobbles. The stop goes a little beyond the invalidation point, so that normal noise doesn't trigger it.
- Size the position. Divide the amount you are willing to lose by the distance to the stop. A wide stop means a small position; a tight stop means a larger one.
If the stop that the chart asks for would need a position too small to be worth it, skip the trade. Moving the stop closer to make the numbers work is the mistake this whole process is built to avoid.
How much noise is normal? A year of data
To see how often a stop gets hit by plain noise, we took one year of 4-hour candles for six large coins on Binance spot, from 20 November 2025 to 19 November 2026 (Binance public market data). At every candle close we placed a test long and a test short, put a stop at a fixed distance, and checked whether any wick reached the stop within the next six candles, which is 24 hours.
These test entries have no edge at all, so the result measures only how far price normally wanders. It is not a strategy test.

The same 2% stop behaved like two different stops. On Bitcoin, normal movement reached it within a day 29% of the time. On Solana and Dogecoin, it was 47%, close to a coin flip. The reason is in the grey numbers: the typical 4-hour range (ATR) was 1.17% on BTC and 1.95% on DOGE, so 2% is a roomy stop on one and a tight one on the other.
The fix: measure the stop in ATR
The Average True Range, introduced by J. Welles Wilder in 1978, measures how much a market typically moves per candle, gaps included. StockCharts' ChartSchool explains the calculation; on TradingView the default is 14 candles. It says nothing about direction, only about size.
When we ran the same test with the stop set as a multiple of each coin's ATR, the six coins lined up almost perfectly:

| Stop distance | Touched within 24 hours (range across six coins) |
|---|---|
| 0.5 × ATR | 70 to 72% |
| 1 × ATR | 47 to 49% |
| 1.5 × ATR | 30 to 32% |
| 2 × ATR | 20 to 22% |
| 3 × ATR | 8 to 10% |
A stop within one ATR of the entry was touched about half the time within a day, on every coin. That is the zone where stops get hit by noise rather than by being wrong. Measured in ATR, a stop means the same thing on Bitcoin and on Dogecoin, which a percentage does not.
One caution: a wider stop is not automatically better. It is hit less often, but each hit costs more, unless you shrink the position to match. Kathryn Kaminski and Andrew Lo showed in the Journal of Financial Markets that if prices moved as a pure random walk, a simple stop-loss rule would always lower expected return; stops can add value when prices have momentum. The stop's real job is to cap the damage when your idea is wrong, not to make a trade profitable.
Four ways to place a stop
1. Structure stop: beyond the last swing
For a long, the stop goes below the most recent swing low that the trade depends on; for a short, above the swing high. If price breaks that point, the up or down move you were betting on has failed. This is the most common method because it ties the stop directly to the chart.
2. Pattern stop: back inside the pattern
For a breakout from a triangle, wedge or range, the idea is that price has left the pattern. If it closes back inside, the breakout has failed. The stop goes on the other side of the pattern's broken line, or beyond the last turning point inside the pattern, with a buffer.
3. Volatility stop: a multiple of ATR
The stop sits a set number of ATRs from the entry, often between 1.5 and 3. A well-known version is the Chandelier Exit, created by Charles Le Beau, which keeps a long position's stop 3 times the 22-period ATR below the highest high of the last 22 periods (ChartSchool). It adapts on its own when the market gets calmer or wilder.
4. Percentage stop: a fixed distance
The stop sits a fixed percentage from the entry, such as 3%. It is the easiest to set, but it ignores both the chart and the coin's volatility, and as the data above shows, the same percentage means very different things on different coins.
| Method | Where the stop goes | Strength | Weakness |
|---|---|---|---|
| Structure | Beyond the last swing low or high | Tied to where the idea fails | Obvious levels attract many stops |
| Pattern | Back inside the pattern | Clear yes or no for breakouts | Large patterns give wide stops |
| ATR | A multiple of ATR from entry | Adapts to each coin's volatility | Ignores the chart's structure |
| Percentage | A fixed % from entry | Simple to set | Too tight on volatile coins |
In practice the best stops combine two of these: a structure or pattern level that says where the idea fails, checked against ATR so the stop is not sitting inside normal noise.
Don't sit exactly on the obvious level
If you can see a level, so can everyone else. Research by Carol Osler at the Federal Reserve Bank of New York, based on the order book of a large currency dealer, found that stop-loss orders cluster just beyond round numbers: sell stops just below them, buy stops just above. Her follow-up work found that price trends moved unusually fast once they reached levels where stops clustered, as one triggered order set off the next (Staff Report 150).
That research is on currencies, not crypto, but the mechanics are the same on any order book. Two habits follow from it:
- Put the stop a real buffer past the level, a fraction of an ATR, not a few dollars.
- Avoid placing it right at a round number, such as a price ending in 000, where many other stops are likely waiting.
Then size the position to the stop
Once the stop is set, the position size follows from one formula: position size = amount you risk ÷ distance to the stop. With a 10,000 USDT account and 1% risk per trade, you can lose 100 USDT:
| Stop distance | Position size | Loss if the stop is hit |
|---|---|---|
| 2% | 5,000 USDT | 100 USDT |
| 4% | 2,500 USDT | 100 USDT |
| 6% | 1,667 USDT | 100 USDT |
A wider stop does not mean more risk. It means a smaller position for the same risk. Fees and slippage add a little to the loss, so count them too; our guide to calculating position size walks through that step by step.
On futures, leverage only changes the margin you post, not this math. It does add one rule: your stop must sit well before your liquidation price. Binance, for example, liquidates on the mark price, and a liquidation can cost more than a planned stop would have.
Stop-market or stop-limit?
How the stop is filled matters as much as where it sits. A stop-limit order turns into a limit order when triggered, and Binance warns that it "may never be filled" if price moves past your limit. A stop-market order exits once triggered, but the fill can land beyond your stop in a fast market. For a protective stop, a fill a little worse than planned is usually better than no fill at all. Our guide to crypto order types covers the differences in detail.
Moving the stop after you enter
Never move a stop further away once the trade is open. It breaks the sizing math: a planned 1% loss quietly becomes 2% or 3%, and the stop that was supposed to protect you turns into a hope.
Tightening a stop is different, as long as the rule was set before the trade. Two common rules:
- Breakeven after the first target. Once part of the position is closed at a profit, the stop on the rest moves to around the entry.
- A trailing stop. The stop follows price at a fixed ATR distance, as the Chandelier Exit does, and never moves back.
What hurts is deciding on the spot. Moving a stop in the middle of a trade is usually a reaction to fear or hope, not to new information.
How Navixa Price Action handles the stop
Navixa Price Action follows the idea, stop, size order on every trade. It enters only after a candle closes beyond a wedge, triangle or channel line, and at that breakout close it sets the stop, two targets and the position size together, risking 2% of the balance per trade by default. After the first target, the stop on the rest of the position moves just past the entry.
In the published Bitcoin backtest (BTCUSDT perpetual, 4-hour candles, October 2019 to October 2026, 0.05% commission and 5 ticks of slippage included), 27% of the 58 positions were stopped out before the first target, and 38% hit the first target and then closed at the moved stop. The largest single loss was 210.97 USDT, which was -1.0R: the planned risk, fees included. Every trade is listed on the Price Action results page. These are past results on one market and do not promise future ones.
FAQ
What is a good stop loss percentage for crypto?
There isn't one that works for every coin. In our one-year test, the typical 4-hour ATR was 1.17% on Bitcoin and 1.95% on Dogecoin, so the same percentage means a roomy stop on one and a tight stop on the other. Set the stop from the chart and check it in ATR, then let the percentage be whatever results.
Should my stop be based on a wick or a candle close?
A stop order on an exchange triggers on price, so a wick can hit it. Some traders use a close-based exit instead and only act when a candle closes beyond the level. That avoids wick noise, but nothing protects you during the candle, so it needs a smaller position or a backup stop further away.
Is stop hunting real?
Stops do cluster around obvious levels, and price can move fast through them as one stop triggers the next. Whether someone pushes price there on purpose is hard to prove. The practical answer is the same either way: keep a buffer past the obvious level and avoid round numbers.
Do I always need a stop loss?
On leveraged trades, yes: without one, the liquidation price becomes your stop, at the worst price. Some spot strategies, such as long-term accumulation without leverage, work without a stop, but then the risk shows up as deep and long drawdowns that you need to be able to sit through.
The bottom line
A good stop starts with the question "where am I wrong?", sits outside the normal noise for that coin, and is paired with a position size that makes hitting it a small, planned loss. A percentage you picked because it felt comfortable does none of those things. To see a strategy that sets the stop, targets and size before every entry, look at Navixa Price Action and its full trade list on the results page.



