Short answer: a stop order is an instruction to get out, not a guarantee of the price you typed. When it triggers, your broker sells at the best price actually available. If price gapped straight past your level, that fill can be a long way below it. The difference is slippage.
The distinction that catches people out
There are two separate prices in a stop order and they are not the same thing:
- The trigger price - the level that turns your order live. You choose this.
- The fill price - what someone was actually willing to trade at when it went live. The market chooses this.
In a calm market they are close enough that the difference never registers. In a fast one there may be no buyer at your level at all, and the order fills at the first price where one exists.
Why gold is worse than most
Two things stack up. Gold moves in large nominal steps, so a violent second covers more ground than it would in a major currency pair. And its worst seconds are scheduled: the release calendar tells everyone in advance when liquidity will thin out.
In the seconds around a release the order book empties. Market makers widen or step away entirely rather than quote into a number they have not seen. Price does not slide through those levels - it reprints somewhere else, and there is nothing in between to fill against.
What it costs, in the only terms that matter
Slippage is not a rounding error on a bad day. It is the difference between the loss you sized for and the loss you take, and it lands on the trade where you were already wrong.
For scale, the worst single leg in our published record is -1,057.6p, and the deepest peak-to-trough fall is -10,809p. Those are outcomes measured at the level, before any gap is added on top - which is exactly why a backtest flatters a strategy that trades through releases.
The scheduled days it happens on
The releases that produce this are known in advance and there are not many of them: the jobs report, the inflation print, and the Fed decision. Dates and the record on every past one are on the news page.
What our own strategy did on NFP days
This is the part no other page can show you: what a specific rules-based gold strategy actually did on these exact dates. It covers 9 verified NFP releases, and it grows by one each time another one happens.
4 September 20263 legs-1,542.7p
6 March 20264 legs-1,805.3p
11 February 20261 leg-526p
Losing days included. Every figure above recomputes from the live trade log and refreshes on its own about once a minute.
What can actually be done about it
Nothing removes slippage. What can change is how much is exposed to it when a release lands - the number of positions open at once, and whether they share a single stop level. Our own approach to that, and what it costs in the backtest, is written up on the risk page.