Placing stops by noise level, not by round pips
Posted: Sat Oct 31, 2026 7:04 pm
A common rule is a stop of 3 pips, because it feels reasonable. The market does not care about round numbers. What matters is how far price typically wanders against a trade that eventually works.
The noise-based method measures the adverse excursion of winning trades: how far did price go against the position before reaching the target? The distribution shows where a stop would cut off winners.
If 90 percent of winners never moved more than 1.8 pips against the entry, a stop at 2 pips captures most of them while limiting losses. If the stop is placed at 1 pip, many winners are stopped out.
The distribution varies with volatility, time of day, and pair. A stop that adapts to the current noise level, for instance a multiple of recent short-term volatility, performs more consistently.
Trade-off analysis plots the stop distance against the win rate and the average loss. The optimum is where expectancy peaks, ideally on a broad plateau.
Basing stops on data rather than on habit often reveals that fixed values are either too tight or too wide.
What does the distribution of adverse excursion look like for the winners in the log?
The noise-based method measures the adverse excursion of winning trades: how far did price go against the position before reaching the target? The distribution shows where a stop would cut off winners.
If 90 percent of winners never moved more than 1.8 pips against the entry, a stop at 2 pips captures most of them while limiting losses. If the stop is placed at 1 pip, many winners are stopped out.
The distribution varies with volatility, time of day, and pair. A stop that adapts to the current noise level, for instance a multiple of recent short-term volatility, performs more consistently.
Trade-off analysis plots the stop distance against the win rate and the average loss. The optimum is where expectancy peaks, ideally on a broad plateau.
Basing stops on data rather than on habit often reveals that fixed values are either too tight or too wide.
What does the distribution of adverse excursion look like for the winners in the log?