Does scalping work on gold?

What scalping means, precisely

Scalping is taking many short trades for small individual profits and relying on volume rather than the size of any single win. Holding periods run from seconds to a few minutes. The definition carries a constraint that is easy to skip past: if the profit target is small, the stop has to be small too, or a single loss erases a long run of wins.

That constraint is the whole subject. Everything below is about what a small stop costs on an instrument that moves like gold.

The spread is flat, so stop optimising it

Across a year of bars on a raw-spread account, the median XAUUSD spread was $0.16. It was identical in 23 of the 24 hours. It exceeded $1.00 in 3 bars out of 70,546 — three.

This is the finding most scalping guides get backwards. There is no cheap window to wait for and no expensive one to avoid, at least not on this account type. What does change by hour is how far gold moves, and that changes the same cost a lot:

The cheapest hour to trade costs 2.3% of the typical move and the dearest costs 7.5% — a 3.3× difference produced almost entirely by range, not by spread. If you want a rule from this, it is “trade when gold is moving”, not “trade when it is cheap”.

The stop is what actually kills it

Gold’s median daily range over the same year was $76.12. Against that, we measured how often a stop of a given size is hit within an hour of entry by ordinary movement — no strategy, no direction, just price doing what it does.

A $2 stop is hit 81% of the time. Not because the trade was wrong — because gold moved $2, which it does constantly. To get the hit rate under half you need roughly a $5 stop, and to get it genuinely out of the noise band you are at $10 or beyond, where 34.2% still get hit.

Now put that back against the spread. A $10 stop is a perfectly reasonable gold stop and a hopeless scalping stop: at a 2:1 reward-to-risk it needs a $20 target, which is not a scalp, it is a swing. Conversely a genuine scalping stop of $2 is a coin flip against noise before the strategy has said anything at all.

That is the trap, and it is a structural one rather than a matter of skill. Gold does not punish scalpers through spread. It punishes them through the stop distance scalping forces them to accept.

What this does and does not prove

This is one symbol, one broker, one year, and one account type. On a standard-spread rather than raw-spread account the spread term is materially larger and the conclusion would shift toward the conventional advice. The noise figures are direction-agnostic by construction: they measure how often price travels a given distance against you, which is a property of the instrument and not of any strategy.

What it does not show is that short-term trading on gold cannot work. It shows that the stop has to sit outside the noise band for the trade to be about your idea rather than about volatility — and once it does, the holding period stops being a scalp. Widening the stop and lengthening the hold is a real answer to this problem. Trading a tighter one and hoping is not.

The underlying data is published in full, with method, at spread and volatility by hour. If you want the sizing arithmetic rather than the argument, the stop loss calculator does it against your own account, and our gold scalping bot page sets out how we handle the constraint above in an automated system.

Common questions