ATR and Exit Rules, Drawn Out
Every lesson so far has been about when to get in. This one is about when to get out.
In the Simulator this corner is easy to miss. Entry indicators each get their own template, but the three exit rules — ATR trailing stop, hard stop-loss, take-profit — exist only as optional filters. With no template of their own, plenty of people never find them.
They also behave differently from every other filter. The rest narrow an existing signal; these three add a closing rule the base strategy didn't have at all.
1. True range: the gap counts too
ATR is built from true range, which isn't simply this bar's high minus its low. It's the largest of three measurements: the bar's own range, the distance from its high to the previous close, and the distance from its low to the previous close.
The last two exist because of gaps. A bar that opens far above the previous close and then trades quietly looks calm if you only measure high to low — but something substantial happened. True range catches it.
Three candidates each bar — this bar's own range, and its high and low measured against the previous close. The largest wins.
- Price (close)
- True range
09/05: high−low 168 · high vs prev close 26 · low vs prev close 142 → the bar's own range wins
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The thick vertical line is this bar's high-low range; the small horizontal tick is the previous close. Below is the resulting true range.
ATR is that true range smoothed the Wilder way — the same smoothing RSI and ADX use.
2. ATR is a distance, not a score
RSI and stochastic are trapped between 0 and 100, which is what lets you draw a line at 70 or 80. ATR isn't. Its units are dollars. An ATR of 800 means a bar has lately been travelling about $800.
So there's no overbought level for ATR, and comparing it across assets is meaningless. Same property MACD has.
It comes out in price units, so it has no overbought level and can't be compared across assets.
- Price (close)
- ATR 14
Average ATR was 443 in the first half of this window and 1,048 in the second. As a share of price that's 0.69% against 1.35% — most of the increase is simply that price got higher, which is why rules built on ATR multiply it rather than compare it to a fixed number.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
ATR being larger later in the window isn't only markets getting wilder — price itself got higher.
This is why every rule built on ATR takes the form of a multiple of it. Using a multiple rather than a fixed dollar amount automatically gives more room when the market is moving and less when it's quiet.
3. The trailing stop only moves one way
Here's the first exit rule. A trailing stop sits a multiple of ATR below the highest close reached since entry, and every time that peak rises the stop rises with it.
The important part is that it only rises. When price falls the stop stays where it is — so the further a trade goes your way, the more of that gain is protected.
The stop sits 3 ATR below the highest close since entry. As that peak rises the stop follows it up — and never comes back down.
- Price (close)
- Peak since entry
- Trailing stop
Entered 08/17, stopped out 08/29 at 20.8% — the ×3 stop stayed out of the way through the advance and closed the trade on the give-back.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The dashed line is the peak close since entry; the solid line is the stop trailing beneath it. The trade closes where price crosses the solid line.
Every exit comparison in this lesson uses one and the same trade: entered on the first close above the previous 30 bars' highest close, which lands on 17 August at $64,200.
If that date looks familiar, it should. It's the same bar the golden cross fires in the Moving Averages lesson, one after the Bollinger squeeze breaks.
4. The multiplier is the decision
A trailing stop really only has one number to choose, and choosing it costs something either way.
Tight allows very little give-back, so gains get banked quickly — and ordinary noise closes the trade before the move is done. Wide sits through the noise but hands back considerably more once a trend genuinely ends.
Tight stops exit early and often; wide ones give back more before they act. There's no setting that avoids both.
- Price (close)
- Peak since entry
- Trailing stop
At ×3.0 the stop fires 08/29, at 20.8% from entry.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
Drag the multiple and watch both the exit date and the result move with it.
5. Three exit rules side by side
Now the same trade under each of the three rules. Identical entry; only the way out changes.
All three measured from the same trade: entered 08/17 at $64,200, on the first close above the previous 30 bars' highest close.
- Price (close)
- Trailing stop (×3 ATR)
Trailing stop (×3 ATR) fired on 08/29, at 20.8% from entry.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
Toggle between them. One trade, three different exit dates and three different results.
They're each protecting something different. The hard stop is a floor under a single trade's loss — in this trade it never fired at all, because price never got 10% below the entry. The take-profit banks a target. The trailing stop tries to stay with an advance while capping the give-back.
They aren't exclusive. All three can be switched on together in the Simulator, in which case whichever condition is met first closes the trade.
6. What a take-profit gives up
The take-profit deserves particular care. It raises win rate, which makes backtest summaries look better — while cutting off exactly the handful of large advances that usually carry a trend-following strategy.
It converts an open-ended move into a fixed one. Sometimes that's the trade you wanted; here it wasn't free.
- Price (close)
- Take-profit (+20%)
A +20% target exits on 08/21. The move went on to reach +27.3% on 09/03 — the target was hit, and the rest of the advance happened without the position.
BTC/USDT · 4h · Binance · 2026-08-03 → 2026-09-05
The circled point is the highest close the move actually reached after entry. A lower target exits sooner and leaves more behind.
The filter's own description in the Simulator carries the same warning: measure it rather than assuming it helps, because judged on win rate alone it almost always looks like an improvement.
7. Common misreadings
- ATR says nothing about direction. It measures the size of movement, and grows the same whether price is rising or falling.
- There's no absolute high or low for ATR. It's in price units, so its meaning depends on the price level and the asset.
- A trailing stop never moves down. Once it has ratcheted up it stays.
- A tight stop isn't the safe choice. It makes losses smaller and being shaken out by ordinary noise more likely.
- A take-profit raises win rate and truncates large gains. Judging it on win rate alone leads you astray.
- The three rules combine; whichever fires first ends the trade.
8. Seeing it run in the Simulator
These three live under the exit-and-risk category of optional filters rather than as templates, and they stack on top of whichever base template you're using.
They differ structurally from the other filters in one way worth knowing. The rest work by narrowing existing buy and sell conditions; these three bypass those conditions and force the exit — deliberately, so that another filter's gate can never suppress a stop.
To compare multiples and target percentages rather than guess, the Sweep tool runs a real backtest across every combination in the ranges you pick and ranks them.
- ATR trailing stop — exits when price falls a multiple of ATR below the highest close since entry.
- Hard stop-loss — exits at a fixed percentage below the entry price. The level never moves.
- Take-profit — exits at a fixed percentage above the entry price.
- ATR volatility filter — a separate, entry-side filter that blocks entries when volatility is below a floor.