Ask ten traders where they put their stop-loss and you will hear "just below support", "fifty pips", or "wherever I'm down one percent". None of those answers asks the only question that matters: how far does this market normally wander on its own? A stop placed inside the market's everyday noise is not protection — it is a scheduled donation.
The Average True Range (ATR) is a volatility gauge built from the most basic fact about a candle: how far price travelled. Each period's true range is the largest of three distances — high minus low, high minus the previous close, or the previous close minus the low — so that overnight gaps are counted as movement too. ATR is simply the average of that number over a lookback window, most commonly 14 periods.
The result is expressed in price units. If Bitcoin's 14-period ATR on the 4-hour chart reads 1,200, a typical 4H candle has recently covered about $1,200 of ground. ATR says nothing about direction — it is pure magnitude. It expands when a market turns violent and contracts when it goes quiet, which is exactly the behaviour of the market regime itself. That adaptiveness is what makes it useful for stops.
A fixed-distance stop treats a calm market and a chaotic one identically. Take a 30-pip stop on EUR/USD. In a quiet market where the 4H ATR is 25 pips, that stop sits beyond a full candle's typical reach — reasonable. In a volatile week where the ATR has swelled to 80 pips, the same 30-pip stop sits well inside a single candle's ordinary wobble. It will be hit by noise, and the direction of your idea will never even be tested.
"Just below support" is closer to correct, because it anchors the stop to structure. But key levels are zones, not lines — and the width of the buffer beyond the zone still has to come from somewhere. Volatility is the honest answer.
The worst variant is the affordability stop: "I can only stand to lose $200, so the stop goes 200 dollars away." The market does not know your pain threshold. How much a loss costs you is a position-sizing decision; where the trade is proven wrong is the market's decision. Confusing the two is how tight stops end up in silly places.
The numbers below are illustrative. Suppose a long on Bitcoin at $60,000 with a 4H ATR of $1,200. A volatility-based rule of 1.5× ATR puts invalidation $1,800 away, at $58,200 — beyond the reach of one ordinary candle, so only genuinely adverse movement can hit it. A first target at 3× ATR sits $3,600 away at $63,600, making the risk-reward 1:2 by construction.
Now watch the rule adapt. In a sleepy market with ATR at $600, the same trade stops out at $59,100 and targets $61,800. In a violent one with ATR at $2,400, the stop drops to $56,400 and the target stretches to $67,200. The distances changed with conditions; the logic and the risk-reward did not.
Dollar risk stays constant through sizing. On a $10,000 account risking 1% ($100): with ATR at $1,200 the stop distance is $1,800 per BTC, so the position is about 0.056 BTC. With ATR at $2,400 the distance doubles to $3,600, so the position halves to about 0.028 BTC. Volatility doubled, size halved, worst-case loss identical.
| Multiplier | Stop (BTC example) | Distance | With a fixed 3× ATR target |
|---|---|---|---|
| 1.0× ATR | 58,800 | $1,200 | R:R 1:3, but the stop sits at one typical candle's reach — whipsaw territory |
| 1.5× ATR | 58,200 | $1,800 | R:R 1:2, buffered beyond ordinary noise |
| 2.0× ATR | 57,600 | $2,400 | R:R 1:1.5 — fewer stop-outs, thinner reward per unit of risk |
| 3.0× ATR | 56,400 | $3,600 | R:R 1:1 — rarely hit, but each loss now equals the target |
Every multiplier is a compromise. Tighten it and you die by a thousand cuts: frequent small losses, plus spreads and fees on every re-entry. Widen it and each individual loss grows — and if the target stays fixed, the risk-reward ratio quietly degrades until even a decent win rate cannot save the expectancy.
There is no universally correct number. A multiplier is a parameter, and parameters have to earn their place on data — which is precisely the kind of claim that honest backtesting and walk-forward validation exist to check. What matters most in practice is consistency: a fixed, tested rule produces results you can measure and improve. Stops improvised trade-by-trade produce only anecdotes.
This is not a theoretical preference. SRA Quant's engine builds every setup it scores the same way: invalidation at 1.5× ATR, first target at 3× ATR, so the risk-reward is fixed near 1:2 before any signal is considered — the published methodology spells out the exact rules, including the consequences that choice has for how scores are calibrated. Outcomes are then judged mechanically against those same levels: stop touched first counts as −1R, no exceptions, and every live result is published on the track record, losses included.
The principle is worth copying even without software: let volatility decide where the exit lives, and let your sizing rule decide what a loss costs. When those two jobs stop fighting each other, stops finally do what they were meant to do — end wrong trades cheaply.
SRA Quant places invalidation at 1.5× ATR on every setup it scores. The level is defined before any alert fires, judged mechanically afterwards, and published win or lose.
SRA Quant provides market analysis and educational content only. Nothing on this page or the platform constitutes financial advice, and past or backtested performance does not guarantee future outcomes. Trading involves substantial risk of loss.