Part 3: Time, stop and size · Chapter 8 of 14 · 8 min
The stop measured in ATR
Entering at random on every candle, a 0.5 ATR stop is touched within 12 hours in 80% of cases, a 1 ATR one in 63%, a 1.5 ATR one in 49%, a 3 ATR one in 21%; within 48 hours, 90%, 82%, 74% and 53%. With a 1.5 ATR stop and a 3 R target, a random entry hits the target in 23% of cases and the stop in 70%; our example plan has 41% targets and 53% stops over 79 trades.
Why the stop is measured in ATR
A fixed stop in dollars is an assumption that ignores the market: $20 is a wide stop when the hourly ATR is $13 (the 10th percentile in our data) and a very tight one when it is $33 (the 90th). Measured in ATR, the stop has the same meaning in every condition: «a typical hour of movement» for each ATR.
Taking 1.5 times the ATR at the last candle of each day, the stop in dollars has a median of $27.9, ranges from $20.1 (10th percentile) to $49.4 (90th), with a minimum of $3.7 on 3 April (a day with an $11 range) and a maximum of $164.5 on 30 January. It is the scale the plan uses for its own stops.
Read the chart in words
Daily gold chart with, below, a pink line that rises and falls a lot: the width of the 1.5 ATR stop over time.
| Stop | With ATR = $18.6 (median) |
|---|---|
| 0.5 ATR | $9.3 |
| 1 ATR | $18.6 |
| 1.5 ATR | $27.9 |
| 2 ATR | $37.2 |
| 3 ATR | $55.8 |
How often the stop is touched
The practical question: if I enter at random, with a stop at k ATR, how often does the price touch it before the idea has time to work? We tried, at every candle of the 4,865 and in both directions (about 9,400 cases per horizon), counting the cases in which the low or the high of a later candle reaches the stop:
| Stop | Touched within 12 candles | Within 24 | Within 48 |
|---|---|---|---|
| 0.5 ATR | 80.2% | 85.6% | 90.3% |
| 1 ATR | 62.9% | 73.0% | 81.6% |
| 1.5 ATR | 49.2% | 62.2% | 73.8% |
| 2 ATR | 37.6% | 52.9% | 66.4% |
| 3 ATR | 21.0% | 36.9% | 52.6% |
Three readings.
- Below 1 ATR the stop is almost always touched by noise: at 0.5 ATR it is touched in 80% of cases within twelve hours, and in 90% within two days. Almost no idea has time to work before that.
- Even at 1.5 ATR the stop is touched in 49% of cases within half a day and 74% within two days: it is the example plan's distance.
- The longer the time, the more the stop is touched: whoever holds a position for days must allow for the fact that, idea aside, the price will almost always touch a short distance.
The square-root-of-time rule (chapter 7) allows a hand estimate: the probability of touching a barrier at distance k with a one-candle deviation σ (here 0.77 ATR, measured) in N candles is about 2 × (1 − Φ(k ÷ (σ√N))). It is fine as an order of magnitude, but it overestimates by 2-9 points (for example 57% against 49% at 1.5 ATR in 12 candles), because fat tails (chapter 6) make the typical case calmer than the average one. It is better to use the measured frequencies.
The base frequency of targets and stops
The example plan uses stops at 1.5 ATR, a 3 R target (4.5 ATR) and a maximum duration of 48 candles. If you entered at random on every candle, in both directions, with the same exit rules (stop first on ambiguous candles):
| Outcome | Random entries (≈9,400) | Example plan (79 trades) |
|---|---|---|
| Stop | 69.7% | 53% (42) |
| Target | 23.0% | 41% (32) |
| Time | 7.3% | 6% (5) |
On 300 random series (one-day blocks) the frequency of the random target is 21.7% (95% of cases between 20.3% and 23.5%). A value of about a quarter is also what you get by hand: with a stop of 1 unit and a target of 3, a random movement without drift reaches the target first 1 time in 4.
The example plan does much better than the base: 41% targets against 23%. With 79 trades the statistical error is about ±11 points (at 95%), so an 18-point gap sits above the error. But book 12 said why it is not enough: this rule is the best of 480, and among 480 rules the best is expected to look very good. It is an «interesting and unproven» result.
How to choose the distance
There is no right stop. The stop is at once the price of a mistake and the price of noise: tight, it reduces the loss but gets touched by noise; wide, it does not get touched but costs more and forces you to reduce size (chapter 9). An orderly way to choose:
- Fix how long you want to hold the position (for example 24 candles).
- Look at how often the stop at k ATR is touched in that time (table above) and choose the distance at which you accept that noise.
- Write the distance in ATR, not in dollars.
- Size the position so that the loss at the stop is the share of the account you decided on (book 11 and chapter 9).
What to do
- Measure the stop in ATR and write it in the plan; convert to dollars at the time of entry.
- Avoid stops below 1 ATR unless your rule justifies them in data: within twelve hours they are touched by noise more than 6 times out of 10.
- Compare the result with the random base (here 23% targets): if it is not far above, the edge is not there.
- Do not forget selection: the best of many rules is above the base by construction.
- Update the ATR at each entry: the stop in dollars changes with the regime (chapter 4).
Procedure in four steps
- Choose the maximum duration of the position.
- From the table read how often the stop at k ATR is touched in that duration.
- Write the distance in ATR and the size that follows from it.
- Compare real outcomes and the random base (stop, target, time).
Exercises
Chapter exercises
1. In the chart the lines are entry ($4,741.68), stop ($4,769.30) and target ($4,658.82) of a sell on 21 April 2026 (UTC time); the line on the left is the 80-candle low ($4,744.78). Which letter is the candle in which the stop is hit?
Read the chart in words
| When | What | Price |
|---|---|---|
| 21 Apr 2026 14:00 UTC | entry | 4,741.68 |
| 21 Apr 2026 14:00 UTC | stop: −1 R (27.6 $) | 4,769.30 |
| 21 Apr 2026 14:00 UTC | target: +3 R | 4,658.82 |
| 21 Apr 2026 06:00 UTC | 80-candle low | 4,744.78 |
| 21 Apr 2026 10:00 UTC | Point A | 4,792.16 |
| 21 Apr 2026 14:00 UTC | Point B | 4,784.00 |
| 22 Apr 2026 04:00 UTC | Point C | 4,760.57 |
| 22 Apr 2026 05:00 UTC | Point D | 4,773.59 |
Times in UTC.
2. With random entries, a 0.5 ATR stop is touched within 12 candles in 80% of cases. What does it mean?
3. For a 1.5 ATR stop and a 3 R target, a random entry hits the target in 23% of cases. The example plan hits it in 41%. What caution is needed?
4. With ATR = $18.6, how wide is a 2 ATR stop?
Have a question or want to share your exercise?
Post in the community, or join the free signals room on Telegram.
Educational content, not financial advice. Trading involves risk.