Trailing stops: what they protect, what they cost

The trailing stop has a flattering reputation: it lets gains run and cuts losses. The first half is true. The second deserves a closer look.

What it actually does

A trailing stop does not follow the price, it follows the lows. As long as the stock climbs by making higher and higher lows, the stop moves up beneath each of them. As soon as the stock breaks the latest low, the position exits.

In other words, it does not protect against a fall: it protects against a change of structure. That is an important distinction, because it explains when this tool fits and when it does not.

What it costs

Three costs, rarely spelled out.

You never exit at the high. By construction, the exit happens after the turn. On a position that has run +30%, giving back 8 or 10 on the way out is normal. It is not a bad setting, it is how the thing works.

A stop that is too tight throws you out on noise. A stock that moves 3% a day will hit a stop placed at 2% for reasons that have nothing to do with your scenario. Hence the use of a volatility measure, a multiple of the ATR for instance, rather than a fixed percentage: the stop breathes with the stock.

A stop that is too wide no longer protects much. At the other extreme, a stop placed so low that it is never hit only delays the admission. The exit happens in the end, a good deal lower.

The trade-off has no universal answer

There is no optimal setting, only a setting consistent with your horizon. A swing over a few days and a trend held for several months do not call for the same width.

Two useful markers:

• Test the setting, do not guess it. The same strategy with a stop at 2× ATR and at 3× ATR gives two very different curves. That is exactly the kind of question a backtest settles in seconds.
• A wider stop forces a smaller position. Widening the stop without reducing the size amounts to increasing your risk while believing you are cutting it.

The order that works

In practice, the sequence that holds up best combines the mechanisms rather than choosing between them:

1. An initial stop below the level that invalidates the scenario.
2. Past +1 R, a trailing stop engages, at a distance from the high.
3. At the first target, half the line is sold.
4. The rest follows, the stop tightening as the trade advances, and never coming back down.

The first step protects the capital, the second locks in part of the run without suffocating the stock, the third secures the gain, the fourth leaves room for the long trend if there is one.

One thing the sequence deliberately does not do is move the stop to the entry price at +1 R. We ran that rule, measured it and dropped it on 11 August 2026: it turns a wide, deliberate stop into a stop at 0% and cuts the trades that were paying. Our Verdicts page carries the numbers.

The real trap

It is not a technical one. It consists of moving the stop down "just this once", because the stock needs a bit of air. A trailing stop never goes down. If it goes down, it is no longer a stop: it is hope with an engineer's name.