Trailing drawdown: what it is, and what it has to survive

The rule in one sentence

A trailing drawdown is a loss limit measured from the highest point your account has ever reached, not from what you started with. It follows your equity up. It never follows it back down.

That single asymmetry is the whole thing. Make money and the bar rises behind you; give some back and the bar stays where your best day left it. It is the reason traders fail challenges while still being up on the month, which reads as unfair right up until you look at how it is defined.

Three limits that are commonly confused

Worked through on a $100,000 account with a 10% limit: a static rule fails you at $90,000 whatever happens in between. Take the account to $110,000 first and a trailing rule now fails you at $99,000 — you are up $9,000 and one ordinary run from being out. The trailing version is strictly harder than the static one, and it gets harder the better you trade, which is the opposite of most people's intuition about risk rules.

What the rule actually has to survive

Every explanation of this rule stops at the arithmetic. The more useful question is how big a move a trailing limit has to absorb, and that needs measurement rather than opinion. So we measured the same shape on gold itself: a running maximum of the daily close across 258 full sessions from 2025-07-31 to 2026-08-04, built from 70,952 five-minute bars, and how far price fell below that running maximum before setting a new one.

Gold closed below its own running high on 218 of 258 sessions — 84.5% of the time. The median distance below the peak was 9.47%. The deepest was 28.49%, or $1,570.89 an ounce, measured to the intraday low — which is the number that matters, because a trailing limit is breached on the touch, not on the close.

Those retracements came in 15 distinct episodes. The median lasted only 2 days. The longest lasted 132 days.

Read the third column. Gold sat 10% or more below its running high on 113 of 258 sessions. Not as a crash — as the ordinary condition of a market that had gone up a great deal and then paused.

The year that produced those numbers

Gold opened the sample at $3,363.12 and closed it at $4,053.76 — up about 21%. On those two numbers alone it looks like a market that simply went up.

It did not go up in a line. It peaked at $5,512.99 on 28 January 2026, then spent the next five months below that level, reaching 28% under it at the end of June. The monthly averages trace the arc plainly: $3,368 in August 2025, $5,022 in February 2026, $4,054 by August. Anyone who took a funded account at the top was trading a market that would not revisit its high for the rest of the sample.

What follows if you are choosing a challenge

• ·A trailing limit is live almost all the time. A static limit only becomes relevant after real losses; a trailing one is anchored to a peak the market spends most of its life beneath.
• ·Your best day sets your worst allowance. Under an equity-trailing rule, a spike you never banked still raises the bar permanently. That argues for taking profit off rather than letting a position run purely for the sake of it.
• ·Duration is not the danger — size is. Our study of gold losing streaks found a three-day run cost 98% of what the nine-day run did. A rule keyed on how long a bad patch lasts is watching the wrong variable.
• ·Size against the tail, not the average. The median retracement here was under 10%; the worst was nearly three times that. A limit that only survives the median is a limit that fails.

It is also why we publish all three of our measured drawdowns rather than the kindest one, and say plainly that two of them would breach a 10% limit. A vendor quoting a single flattering drawdown figure is telling you about their best sample, not about your risk.

How to cite this

Published under CC BY 4.0, which permits reuse — including commercial reuse — on the single condition that it is attributed. If you are quoting a figure from this page in an article, a paper or a model answer, this is the attribution:

Tech Kick (2026). "XAUUSD retracement from a running high, 2025-2026"
Derived from 70,952 M5 bars across 258 full sessions.
Licence: CC BY 4.0. DOI: 10.5281/zenodo.21973215
https://techkick.me/blog/trailing-drawdown

Or as BibTeX:

@misc{techkicktrailingdrawdown,
  author       = {Tech Kick},
  title        = {XAUUSD retracement from a running high, 2025-2026},
  year         = {2026},
  doi          = {10.5281/zenodo.21973215},
  howpublished = {\url{https://techkick.me/blog/trailing-drawdown}}
}

The headline figures: gold closed below its own running high on 84.5% of 258 sessions; median retracement 9.47%; deepest 28.49%; 15 retracement episodes, longest 132 days. Underlying data CC BY 4.0 under the DOI above.

What this does not support

One instrument, one year, one broker feed. Fifteen retracement episodes is a small sample, and the single 132-day episode is exactly one observation — it shows such a stretch is possible, not how often to expect one. A different year could contain none.

The dollar figures also belong to a specific price level and do not transfer to a very different one. And close-to-close direction says nothing about what happened inside a day, which matters a great deal to a stop and not at all to this measurement.

The underlying data is published under a DOI and the script that produced these figures is in our bot repository, so the arithmetic can be checked rather than believed. If you find an error, tell us and the correction will appear here with the same prominence as everything else.