Partial fills on XAUUSD: how I log them and when I cancel the rest

[LondonScalper]

Gold execution note — partial fills, not a gold call.

On XAUUSD I still see tickets that fill in pieces, especially around the London–NY handoff or when the tape is thin. Treating a partial as “the full idea” has cost me more than the spread ever did.

My logging fields
• Intended size vs filled size
• Time gap between slices
• Mid at first fill vs mid at last fill
• Whether I cancelled the residual or let it work

Decision rule I actually use
If less than half the intended size is filled and price has already moved through my invalidation, I cancel the rest and manage what I have as a reduced-size trade — or flatten if the thesis is gone. I do not average in to “complete” the ticket. Completing size after the move is a new decision, not the original one.

If the partial is tiny noise (a few percent) and the level still holds, I may leave a working residual with a hard time stop. That time stop is pre-set; I don’t extend it because I’m annoyed.

Weekly review: I sort the journal by partial = yes and ask whether those tickets had worse expectancy than clean fills. Usually they do when I chase the remainder. Curious how others tag partials so the review isn’t just vibes.

[PropScalpDesk]

Partial fills: cancel and log, do not spam

On XAUUSD especially, partial fills are normal under speed. My rule is dull and it has saved me from worse average prices more times than clever order tricks ever did.

• If a partial prints and the remainder hangs: cancel the rest after a short fixed wait in seconds, pre-set before the session.
• Do not fire repeated market orders to “complete the size.”
• Log partial rate by session — if it spikes, size down or stand aside.
• Manage the filled piece with the original invalidation, or scratch if the idea required full size to make sense after costs.

Chasing remainder is how a clean A-setup becomes a cost pile. From a Frankfurt desk, gold speed without a cancel rule is just volunteering for slippage.

I also note whether partials cluster on one broker path or one VPS route. If partial rate jumps after an infrastructure change, I treat it as a path problem before I blame gold itself. Measure first, then adjust size.

How long do you wait on a working remainder before cancel — and is it a written number, or a feeling in the moment?