Page 1 of 1
Broker comparison: XAUUSD average spread during London killzone
Posted: Tue Sep 22, 2026 10:22 am
by LondonScalper
Gold into the London killzone is where my cost sheet either justifies the trade or quietly kills it.
I keep a simple comparison row for XAUUSD: average spread and a rough slippage hint across the accounts I actually use, tagged for the first 90 minutes of London cash. Marketing screenshots do not sit in that sheet. Only my timestamps do.
How I read the killzone cost
1. Spread at decision, not the overnight fantasy print.
2. Fill distance on market orders under 0.2R targets — gold eats small edges fast.
3. Whether the "raw" label still looks raw when the metal starts running.
If broker A is 0.15 cheaper on a quiet open but blows out on the first impulse, it loses the week. Killzone average without a volatility tag is theatre.
I am curious how others build a gold-only cost comparison for London — do you weight the open heavier than the rest of the session, or keep one flat average and accept the noise?
I keep killzone rows separate from all-day gold average because the open is where my scalps live. If a broker only looks cheap in the quiet afternoon, it does not win the comparison for my desk. A short note on whether partial fills showed up that week sits next to the spread number.
Re: Broker comparison: XAUUSD average spread during London killzone
Posted: Sun Sep 27, 2026 4:09 pm
by PropScalpDesk
Rather than weighting the open, I'd stop using the average for gold altogether. What hurts at the London open is the spread when you actually enter or get stopped, and that happens disproportionately when the spread is widest. The 90th percentile over the first 15 minutes tells you more about a venue than the mean over 90.
The practical difficulty is getting the numbers. MT5 doesn't record the spread at the moment of your fill, so I run a small logger that writes bid, ask and time on every tick for XAUUSD between 07:55 and 09:30 London. After a couple of weeks you can compare venues on exactly the minutes you trade.
Re: Broker comparison: XAUUSD average spread during London killzone
Posted: Sat Oct 17, 2026 1:53 am
by Shadow Trader
Splitting the percentile by trade side changed my gold numbers more than switching from the mean to p90 did. For a long, the stop is triggered on the bid, so a spread blowout barely touches it. For a short, the stop closes on the ask, and a widening spread can take you out while the bid never moves.
So I keep two columns for gold at the open. One is the p90 spread in the 5 minutes after my entries, the other is the p90 in the 5 minutes around stop outs on shorts, and on one account the second number was nearly double the first.
That explained a run of short stops I had been filing as bad luck. The logger PropScalpDesk describes gets heavy quickly, so mine writes one row per second with the widest spread seen in that second, which keeps the file small and the spikes intact.
I don't weight the open at all. I divide the p90 spread by my usual target, and on a 2.00 dollar target a 40 cent p90 means a fifth of the move is gone before any slippage.
Has anyone seen a venue whose p90 stays tight on the first impulse but whose fills still slip? That combination would change how I read the spread column.
Re: Broker comparison: XAUUSD average spread during London killzone
Posted: Sun Oct 18, 2026 2:43 am
by Shadow Trader
One thing I would watch with a couple of weeks of logging is how few mornings actually set the 90th percentile. Ten sessions of 15 minutes is only 150 minutes of data, so the worst 15 minutes can all come from one bad open, and a venue's p90 can really be a single data surprise.
I found that out when a venue I had rated badly looked fine the following month. Its bad fortnight included the morning gold dropped 20 dollars before 08:00 London, and that one session was most of its p90.
So I now want at least 30 mornings before ranking anything, and I look at the p90 again with the single worst session removed. If the ranking flips when one day is dropped, the comparison is not finished yet.
The two loggers also need the same clock. One of mine was stamped in server time two hours ahead of London, and for a while I was comparing 08:00 on one account with 06:00 on the other.
Does your logger take its time from the platform or from your own machine, PropScalpDesk? Mine now writes both.
Re: Broker comparison: XAUUSD average spread during London killzone
Posted: Fri Oct 30, 2026 12:44 am
by Shadow Trader
Psychological levels like 1.2000 and 1.2050 are old news. The reason they keep getting mentioned is that they keep influencing behaviour, even in an era of algorithms.
Several things stack up. Option strikes often sit at round numbers. Stop orders and take-profits cluster there because humans find them easy to pick. Banks place client orders at them. All of that creates real liquidity events when price approaches.
The catch is that the reaction is not constant. A round number near a major high may produce a strong bounce. A round number in the middle of a trending move may produce nothing at all. And a level that held ten times can fail on the eleventh in a burst that wipes out every tight stop behind it.
A more disciplined approach is to track reactions. For one month, log each touch of a round number on one pair, and record whether price bounced, stalled, or broke through, along with the session and context.
It also helps to separate the big figures, like 1.2000, from the half-figures and quarter-figures. Big figures generally attract more reaction than the in-between levels, and a log that distinguishes them often shows a clear difference in how reliably price responds.
What fraction of round-number touches on your pair actually produced a tradable reaction, and what separated those from the others?
Re: Broker comparison: XAUUSD average spread during London killzone
Posted: Fri Oct 30, 2026 6:08 am
by Shadow Trader
Most traders watch price. A few also watch spread over time, and the picture it gives is revealing.
Plot the spread by hour of day for a pair, averaged over a month. A typical curve sits low during London and New York overlap, rises during the Asian session, and spikes at rollover. Specific events appear as sharp peaks.
That curve is a map of when trading costs the least. Scalpers can use it to choose trading windows, avoiding hours where the cost ratio is poor.
It also reveals broker behaviour. A broker that widens spreads more than others during certain hours, or shows sudden jumps at odd times, leaves a signature in the data.
Many platforms can log the spread through a simple script or indicator, and exporting the data takes minutes. A month of logging builds a useful dataset.
What does the spread curve look like on the main pair, and does it match the hours traded?