Sharing a process, not a broker advert.
I track AUDUSD spread around a fixed clock stamp (broker time aligned to my desk) so “AUD is fine” isn’t a vibe. One of the stamps I care about is ~13:30 London — late morning / into midday depending on season — because my AUD attention window sits there more than at the open chaos.
What I log
• Median spread in a ±5–10 min window
• Worst print in that window
• Session notes (news day vs quiet)
If your AUDUSD cost spikes later than the open, you can waste a clean mid-session setup on a pair that’s no longer cheap enough for the target.
Anyone else logging fixed-time spread samples (not just “Asian vs London” averages)? Curious how AUDUSD behaves on your feed mid/late London vs EURUSD.
Not advice — cost logging.
AUDUSD average spread at 13:30 London time from my log
-
LondonScalper
- Posts: 965
- Joined: Sat Sep 05, 2026 7:54 am
-
PropScalpDesk
- Posts: 521
- Joined: Sat Sep 19, 2026 7:50 pm
Re: AUDUSD average spread at 13:30 London time from my log
Clock-stamped AUDUSD spreads beat vibes
Logging median spread at a fixed stamp is the right habit. “AUD is fine” is how I used to take late-morning tickets that were already dead after costs. Around 13:30 London my Frankfurt attention on AUDUSD is selective; the sheet decides if the window is open.
What I mirror in my own log: median, 90th percentile, and a simple “tradeable / not” flag against my minimum edge in points. One quiet Tuesday does not rewrite the distribution. I want a month of stamps before I change pair priority.
Rule: if the stamp’s median plus commission eats more than a defined fraction of my average winner, AUDUSD is observation-only for that window — no “just a small one.”
I align broker server time to desk time once and note DST shifts so the stamp does not silently drift. A wrong clock turns a careful sample into folklore.
Does your 13:30 sample still look tradeable in summer vs winter liquidity, or do you seasonalise the pair off the sheet?
Logging median spread at a fixed stamp is the right habit. “AUD is fine” is how I used to take late-morning tickets that were already dead after costs. Around 13:30 London my Frankfurt attention on AUDUSD is selective; the sheet decides if the window is open.
What I mirror in my own log: median, 90th percentile, and a simple “tradeable / not” flag against my minimum edge in points. One quiet Tuesday does not rewrite the distribution. I want a month of stamps before I change pair priority.
Rule: if the stamp’s median plus commission eats more than a defined fraction of my average winner, AUDUSD is observation-only for that window — no “just a small one.”
I align broker server time to desk time once and note DST shifts so the stamp does not silently drift. A wrong clock turns a careful sample into folklore.
Does your 13:30 sample still look tradeable in summer vs winter liquidity, or do you seasonalise the pair off the sheet?
Re: AUDUSD average spread at 13:30 London time from my log
One thing to check in that sample: 13:30 London is exactly when 08:30 New York data comes out for nearly the whole year. CPI, NFP, retail sales, and weekly jobless claims every Thursday. On those days the AUDUSD spread around 13:30 is a news spread, and on quiet days it's a normal midday spread, so one median across both mixes two different distributions. Tagging each stamp as data day or not and looking at them separately would make it much cleaner, or the stamp could move to 13:15 so it sits before the release. Also, from 26 to 30 October the UK has changed clocks and the US hasn't, so the US releases land at 12:30 London and your 13:30 reading becomes the hour after the print. That week is easy to mislabel later.
It’s Fairman 
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: AUDUSD average spread at 13:30 London time from my log
Splitting the stamps by data day, as Fairman suggests, removes more of the sample than people expect. Weekly claims alone make every Thursday a data day at that time, and once CPI, payrolls, retail sales and PPI are added it was close to a third of my stamps in a normal month.
That leaves about 14 clean stamps a month. PropScalpDesk wants a month before changing pair priority, and I would stretch that to two for the quiet bucket alone.
The worst print field is the one I changed. A single tick at 3 pips in a ten minute window tells you little. Now I log how many seconds of the window the spread spent above my threshold.
For the cost rule, writing the fraction down helped. My average AUDUSD winner is about 6 pips, and with a 0.9 median spread plus 0.6 commission the cost is 25 percent of it, which is over the 20 percent line I use, so most quiet 13:30 stamps fail for me anyway.
Does anyone keep the stamp on data days at all? I have started treating those as a separate news log, since what I want from them is the recovery time back to the quiet median rather than the spread at the print.
That leaves about 14 clean stamps a month. PropScalpDesk wants a month before changing pair priority, and I would stretch that to two for the quiet bucket alone.
The worst print field is the one I changed. A single tick at 3 pips in a ten minute window tells you little. Now I log how many seconds of the window the spread spent above my threshold.
For the cost rule, writing the fraction down helped. My average AUDUSD winner is about 6 pips, and with a 0.9 median spread plus 0.6 commission the cost is 25 percent of it, which is over the 20 percent line I use, so most quiet 13:30 stamps fail for me anyway.
Does anyone keep the stamp on data days at all? I have started treating those as a separate news log, since what I want from them is the recovery time back to the quiet median rather than the spread at the print.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: AUDUSD average spread at 13:30 London time from my log
I now have a few weeks of the separate news log, and the recovery times fall into two groups. On weekly claims days the AUDUSD spread was back to the quiet median within about 90 seconds, while CPI and payrolls days took between 4 and 9 minutes.
That suggests claims days need not leave the clean bucket entirely. If the window starts at 13:35 rather than straddling the print, most Thursdays become usable again and the monthly sample grows back towards 18.
On the summer versus winter question PropScalpDesk raised, my winter stamps have been tighter by about 0.1 pip at the median. At 13:30 London it is 00:30 in Sydney in winter and 22:30 in July, so either way the Australian side has gone home and the pair runs on London and New York flow.
The late October week Fairman flagged is now on my calendar. For 26 to 30 October I will log a 12:30 stamp as the data stamp and keep 13:30 as a quiet one, tagged so the week can be dropped if it looks odd.
Has anyone compared the same stamp on AUDUSD and NZDUSD? I would expect the kiwi to be worse at that hour, but I have no numbers yet.
That suggests claims days need not leave the clean bucket entirely. If the window starts at 13:35 rather than straddling the print, most Thursdays become usable again and the monthly sample grows back towards 18.
On the summer versus winter question PropScalpDesk raised, my winter stamps have been tighter by about 0.1 pip at the median. At 13:30 London it is 00:30 in Sydney in winter and 22:30 in July, so either way the Australian side has gone home and the pair runs on London and New York flow.
The late October week Fairman flagged is now on my calendar. For 26 to 30 October I will log a 12:30 stamp as the data stamp and keep 13:30 as a quiet one, tagged so the week can be dropped if it looks odd.
Has anyone compared the same stamp on AUDUSD and NZDUSD? I would expect the kiwi to be worse at that hour, but I have no numbers yet.
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: AUDUSD average spread at 13:30 London time from my log
A strategy's average result per trade is a single number. How reliable is it? The bootstrap offers a way to find out without heavy theory.
The method: take the list of trade results, draw randomly with replacement to build a new list of the same length, and compute the average. Repeat a few thousand times. The spread of those averages shows how much the result could vary by chance. The 2.5th and 97.5th percentiles form a 95% confidence interval.
If that interval comfortably excludes zero, the edge is probably real in the statistical sense. If it spans zero, the result could be luck.
The technique makes few assumptions about the distribution of trade results, which matters because trading returns are skewed and fat-tailed. It also works for any statistic: expectancy, profit factor, maximum drawdown.
A caveat: the bootstrap assumes trades are independent and drawn from a stable process. Clustered wins and regime shifts violate that, so the interval may be too optimistic.
Still, a ten-line script, or a spreadsheet macro, delivers a far more honest picture than a single average.
A hands-on version in a spreadsheet: put the trade results in a column, use a formula to pick random rows, compute the mean, and fill the formula down a few thousand times. Then take percentiles of the resulting column. No special software required, and the exercise builds intuition about how much a single average can wander.
What does the bootstrap interval look like for the personal expectancy, and does it include zero?
The method: take the list of trade results, draw randomly with replacement to build a new list of the same length, and compute the average. Repeat a few thousand times. The spread of those averages shows how much the result could vary by chance. The 2.5th and 97.5th percentiles form a 95% confidence interval.
If that interval comfortably excludes zero, the edge is probably real in the statistical sense. If it spans zero, the result could be luck.
The technique makes few assumptions about the distribution of trade results, which matters because trading returns are skewed and fat-tailed. It also works for any statistic: expectancy, profit factor, maximum drawdown.
A caveat: the bootstrap assumes trades are independent and drawn from a stable process. Clustered wins and regime shifts violate that, so the interval may be too optimistic.
Still, a ten-line script, or a spreadsheet macro, delivers a far more honest picture than a single average.
A hands-on version in a spreadsheet: put the trade results in a column, use a formula to pick random rows, compute the mean, and fill the formula down a few thousand times. Then take percentiles of the resulting column. No special software required, and the exercise builds intuition about how much a single average can wander.
What does the bootstrap interval look like for the personal expectancy, and does it include zero?