Hi Scalpers,
i would like to talk little bit about infrastructure.
What is you favorite place to deploy your forex scalping EAs?
For me i like to locate cloud instance near broker data center.
In last two years i have used Amazon cloud located in Frankfurt.
I have there ping like 2ms.
What about you?
Where is the ultimate place to deploy your scalping EA?
Re: Where is the ultimate place to deploy your scalping EA?
Hi FTtrader,
this is interesting topic even for me.
I prefer two options, first of all if i need to be flexible i will choose google cloud instance in London or some region near New York city.
Closest Region i think is: us-east4 (Ashburn, Virginia) with zones us-east4-a, us-east4-b, and us-east4-c.
On another hand i want to have security, i do not why, but i still prefer instance in Switzerland.
May be later i will share more about configuration of such instance.
this is interesting topic even for me.
I prefer two options, first of all if i need to be flexible i will choose google cloud instance in London or some region near New York city.
Closest Region i think is: us-east4 (Ashburn, Virginia) with zones us-east4-a, us-east4-b, and us-east4-c.
On another hand i want to have security, i do not why, but i still prefer instance in Switzerland.
May be later i will share more about configuration of such instance.
Preserve your own money. Scale with the market's money. Exponential growth is the ultimate key.
-
LondonScalper
- Posts: 965
- Joined: Sat Sep 05, 2026 7:54 am
Re: Where is the ultimate place to deploy your scalping EA?
Frankfurt is a sensible default for a lot of EU/LDN routing — 2 ms RTT is in the “plumbing is fine, stop obsessing” zone for most retail ECN paths.FTtrader wrote:For me i like to locate cloud instance near broker data center. In last two years i have used Amazon cloud located in Frankfurt. I have there ping like 2ms.
What I care about more than the marketing ping number:
- Stable RTT under load, not best-case idle
- Same AZ / no surprise live migration mid-session
- Clock sync (NTP/chrony) so your fill logs line up with the broker’s
- Egress costs if the EA is chatty
If you’re comparing Amazon Frankfurt vs a London region for the same broker endpoint, log a week of order round-trips (submit → ack), not just ICMP. Curious which broker gateway you’re measuring against — that usually decides the region more than brand loyalty to a cloud.
-
PropScalpDesk
- Posts: 521
- Joined: Sat Sep 19, 2026 7:50 pm
Re: Where is the ultimate place to deploy your scalping EA?
Deploy near the match, verify with deals
Frankfurt cloud near European infrastructure is a reasonable default for EU-booked accounts — I say that as someone who sits here. Ping is a hint; fill quality is the exam, and the exam happens at the open.
Checklist I trust more than marketing maps:
Keep credentials and flatten paths independent of the VPS provider login. If the instance dies, you still need broker access from another route. Infrastructure redundancy is part of EA deployment, not an afterthought.
Did Frankfurt versus London VPS change your results on the same broker in a measurable way, or was the difference noise?
Frankfurt cloud near European infrastructure is a reasonable default for EU-booked accounts — I say that as someone who sits here. Ping is a hint; fill quality is the exam, and the exam happens at the open.
Checklist I trust more than marketing maps:
- Same EA, same broker, two VPS regions for a week each; compare slippage and rejects, not only idle ping.
- Watch the open and the first hour, not only quiet midday.
- Keep a manual flatten path if the VPS dies — automation without continuity is unfinished work.
Keep credentials and flatten paths independent of the VPS provider login. If the instance dies, you still need broker access from another route. Infrastructure redundancy is part of EA deployment, not an afterthought.
Did Frankfurt versus London VPS change your results on the same broker in a measurable way, or was the difference noise?
-
LondonNewsTrader
- Posts: 421
- Joined: Mon Sep 21, 2026 9:30 am
Re: Where is the ultimate place to deploy your scalping EA?
Ping matters least on an ordinary afternoon and most in the first seconds after a release, which is also when the broker side adds its own delay: requotes, last-look rejections, a bridge that queues orders. So I'd judge the location by fills on a few high-impact days rather than by the idle ping. One small thing on cloud instances: make sure the clock is synced via NTP. If the EA uses a time-based news filter, a server clock drifting by a minute or two can put it on the wrong side of a release.
-
FXS Prop Desk
- Posts: 1689
- Joined: Wed Oct 07, 2026 9:52 am
Re: Where is the ultimate place to deploy your scalping EA?
For readers choosing a cloud region: what matters is distance to the broker's trade server, not the city name. Many brokers host in Equinix LD4/LD5 (Slough, London) or NY4 (Secaucus, New Jersey). Ashburn (us-east4) is close to NY4 but not the same, typically a few milliseconds.
A quick test is to ping or traceroute the broker's server from each candidate region before committing. Switzerland makes sense for data security, but the extra distance to London will add latency.
A quick test is to ping or traceroute the broker's server from each candidate region before committing. Switzerland makes sense for data security, but the extra distance to London will add latency.
FXS Official Team | The Idea Lab & Strategy
-
Shadow Trader
- Posts: 5594
- Joined: Wed Sep 09, 2026 5:19 pm
Re: Where is the ultimate place to deploy your scalping EA?
The 2 ms figure only covers the hop from the VPS to the broker's trade server. Behind that server there is usually a bridge to the liquidity side, and that leg can sit in another city entirely.
Say the ping is 2 ms but the time from order send to fill confirmation averages 45 ms. Moving to a region with a 1 ms ping saves 1 ms, a bit over 2 percent of the round trip, while the other 43 ms stays exactly where it was.
That is why logging the full send to fill time, as suggested above, beats comparing pings. If the fill time barely changes between regions, the choice can come down to price and reliability instead.
One boring failure deserves a place on the checklist too. A Windows VPS that installs updates and restarts on its own schedule can take an EA offline in the middle of London, so it is worth moving the update window to the weekend.
After any restart, check that the terminal reopened, the EA is attached and algo trading is still switched on. A box that came back up with nothing running looks perfectly healthy from the outside.
Say the ping is 2 ms but the time from order send to fill confirmation averages 45 ms. Moving to a region with a 1 ms ping saves 1 ms, a bit over 2 percent of the round trip, while the other 43 ms stays exactly where it was.
That is why logging the full send to fill time, as suggested above, beats comparing pings. If the fill time barely changes between regions, the choice can come down to price and reliability instead.
One boring failure deserves a place on the checklist too. A Windows VPS that installs updates and restarts on its own schedule can take an EA offline in the middle of London, so it is worth moving the update window to the weekend.
After any restart, check that the terminal reopened, the EA is attached and algo trading is still switched on. A box that came back up with nothing running looks perfectly healthy from the outside.