OptiNod Academy
Trailing stops — Changing the width changes the win rate and the profit
The trailing width changes both the exit timing and the average result. This article compares 2, 3 and 4×ATR and uses real pullback cases to cover what to check when choosing a width.
A trailing stop is a rule that raises the stop as price follows. When choosing the width, look not only at the win rate but also at the difference between trades that ended early and trades held for a long time.
Different trailing widths were applied to one BTC trade entered on November 7, 2024. The 2×ATR ended at +3.40R on November 25, the 3×ATR at +4.10R on December 19 and the 4×ATR at +3.33R on December 20. The widest setting exited last, but it did not earn the most.
A trailing stop moves the stop up as price moves favorably, and exits when a pullback reaches the stop. This calculation takes the level an ATR multiple below the highest price since entry as the candidate stop. A stop that has been raised is never lowered again.
The common entry and initial stop are the same as in the previous parts. The highest price and ATR are updated after each daily close is confirmed, and the new stop applies from the next bar. The initial stop is fixed at 2×ATR, and only the trailing width was varied. These are results from the BTC daily sample, excluding commissions and slippage.
A narrow width can cut losses but can also end a rise early
In this sample the average loss was -0.60R for 2×ATR and -0.76R for 3×ATR. With a narrow width the stop sits closer to price sooner, so many trades ended with a small loss before reaching the initial stop.
On the other hand, the average of winning trades was +1.71R and +2.86R respectively. The trade entered on July 26, 2021 shows the difference. The 2×ATR ended at +0.29R on the next day's pullback, but the 3×ATR held through the later rise and ended at +3.04R on September 7. In exchange for more chances to end with a small loss, there are also trades that exit mid-trend.
You cannot conclude that the win rate moves in one direction with the width. The win rate was 43.0% for 2×ATR, 37.7% for 3×ATR and 40.9% for 4×ATR. Holding periods differ, so the signals actually received also differ, and this should not be read as a rule that a narrower width raises the win rate.
A wide width's good performance can depend on a few trades
The average result per trade was +0.39R for 2×ATR, +0.61R for 3×ATR and +2.08R for 4×ATR. But the 4×ATR result was heavily driven by a +39.7R trade entered on March 30, 2019 and held until June 27. This single trade accounts for about 43% of total profit.
Excluding that trade, the 4×ATR average is +1.21R. It is still positive, but it differs from the impression the +2.08R figure gives. For 3×ATR as well, the top 5 trades accounted for about 86% of total profit. In a strategy that holds long trends, such large trades can have a big effect on performance.
When choosing a width, check the contribution of the top trades along with the average. This does not mean the result without the large trades is the strategy's "true performance". It does help you check what results you would face when similar trends become scarce.
Past pullbacks are material for choosing a width, not an upper limit on the next pullback
The pullback from the November 22, 2024 high to the November 26 low was 2.53 times the ATR at the time of the high. From the December 17 high to the December 20 low it was 4.09 times. Even within the same advance, the size of the pullbacks differed.
Setting the width wider than the recent pullback does not guarantee that the position will survive the next one. Also, the trailing stop in this calculation updates ATR every bar and holds the stop after raising it, so the pullback width measured with the ATR at the high does not match the actual stop path exactly. After choosing candidate widths, you need to check how the stop moved bar by bar.
When you apply a width chosen on daily bars to another timeframe, compare again as well. On smaller timeframes the makeup of trends and pullbacks and the share of trading costs can differ. Do not take the earlier results as a guarantee of performance for the same multiple.
The initial stop and the trail width play different roles
The initial stop sets the loss you accept right after entry. The trail width sets how closely the stop follows after price has moved favorably. The two values do not need to be equal.
Simplifying by assuming ATR is constant, when both the initial stop and the trail width are 2×ATR, the stop can rise as much as the highest price rises above the entry price. If the trail width is 3×ATR, the highest price first has to rise by 1×ATR before the candidate stop equals the initial stop. In the actual calculation ATR also changes, so it has to be checked every bar.
We look at the initial stop mainly through MAE and stop placement, and the trailing stop mainly through pullbacks and the path after the exit. If you change both at once it is hard to tell what produced the performance difference, so compare one at a time first.
Changing the exit tool changes how the stop follows price
The rule in this part is similar to the Chandelier Exit, which uses the highest price and ATR as its basis, but here it uses the highest price since entry. The result can differ from implementations that use the recent highest price over a fixed length.
Parabolic SAR moves its line according to the extreme point and the acceleration factor, and Supertrend updates an ATR band around the bar's midpoint. The reference price and the conditions for keeping the previous line differ, so you cannot say that either tool's stop is always closer or lower.
The calculation methods can also be checked in the official SAR description and the official Supertrend description. Compare by matching the actual calculation rules rather than the tool names. Even with the same ATR multiple, it becomes a different strategy depending on the observation range of the highest price, whether the line can be lowered, and when the updated line is applied.
When comparing, fix the moment the stop takes effect
- Initial stop: Place it two times the prior bar's ATR(14) below the entry price. This distance is 1R.
- Update: After the close is confirmed, compute the candidate by subtracting 3×ATR from the highest price since entry.
- Hold: From the next bar, apply the highest of the initial stop, the previous stop and the new candidate.
- Fill: The assumption is that if the open is below the stop, the exit is at the open, and otherwise it is at the stop price when touched.
- Comparison: Look at the win rate, average profit, average loss, holding period and the contribution of large trades over the same period.
When narrowing the width, check the trades that ended early, and when widening it, check the profit given back in pullbacks. After judging whether that difference fits your own holding purpose, check the same rule in a period that was not used in the comparison.