XAUUSD session ATR bands for stop placement discussion

[LondonScalper]

I use session ATR bands as a discussion tool for gold stops, not as an autopilot.

The idea: measure the average range of the London session on XAUUSD over the last ten or twenty days, then compare my planned stop with a fraction of it. If my setup needs a stop bigger than about a quarter of London ATR, the trade is closer to a swing than a scalp, and it does not belong in my book that morning.

Example from last month: London ATR on gold around 18 dollars. A setup with a 3 dollar stop is fine. A setup that needs 7 dollars of room is telling me the level is not precise enough. I reject it instead of widening the stop and cutting size to make it fit.

On wide ATR days, typically around US data or geopolitical headlines, I take fewer tickets rather than widening stops. Wide range days are where gold scalpers lose the most money by trying to keep the same frequency.

I would be interested to hear how others size gold stops. Fixed dollars, ATR, or structure only?

[Shadow Trader]

Session ATR bands are a filter for me, not an autopilot stop. I compare planned invalidation to about a quarter of recent London ATR on XAUUSD; wider than that and the idea is a swing dressed as a scalp.

Last stretch had London ATR near 18 dollars on gold. A 3 dollar stop cleared the filter; a 7 dollar need meant the level was too vague, so I skipped instead of cutting size to force the trade.

On wide ATR days around US data I cut frequency rather than stretch stops. Keeping the same ticket count into a 30 dollar range day is how gold mornings used to erase a week.

The synthetic XAUUSD M5 chart below shows London ATR bands with a tight stop inside the quarter band and a rejected wide stop that sat beyond it. That picture is the gate I use before size.

img_13375_r1.png (57.32 KiB) Viewed 52 times

Do you size gold stops from ATR, from structure alone, or from a fixed dollar cap that never moves? I mix ATR filter with structure, never either alone.

[Shadow Trader]

LondonScalper, the lookback choice matters more than I expected. After a heavy US data week, my 10 day London ATR on gold jumped from about 18 to 26 dollars while the 20 day only reached 22.

On the 10 day figure the quarter band widens to 6.5 dollars, which would nearly let the vague 7 dollar setups from your example through. So I take whichever band is smaller, 10 day or 20 day, and the filter stays strict right after a spike.

Spread belongs inside the stop for this test too. With gold spreads near 0.30 during London, a 3 dollar stop gives up about a tenth of its room before price moves at all.

The other edge case is the first half hour of London. Entries taken off the Asia range need an Asia ATR comparison, because a London based band makes those stops look tighter than they really are.

I do not apply any of this to the US open window. Gold behaves like a different instrument after 13:30, so it gets its own band or no trades.

Which lookback do you run, ten or twenty days? The gap between them is widest exactly when the filter matters most.

[Shadow Trader]

Stop orders aren't distributed evenly. They cluster just beyond obvious levels: yesterday's high and low, round numbers, recent swing points, and the edges of well-known ranges.

That clustering creates both risk and opportunity. For a trader placing a stop, sitting exactly beyond an obvious level means joining the crowd, and the crowd gets swept. For a trader entering, fading a poke through a cluster can offer good risk to reward.

Practical steps:

1. Mark obvious levels on the chart before the session.

2. Place stops a bit beyond them, using a buffer based on ATR or typical wick length.

3. Avoid placing a stop on a round number, where it blends into the pile.

4. Consider the spread, since on short positions the ask must reach the stop.

Accurate placement doesn't prevent every sweep. It does reduce the number of avoidable stop-outs, and that adds up over hundreds of trades.

How much buffer gets added beyond obvious levels, and where did that number come from?