MT5 Broker Ping from Amsterdam: Forex VPS Latency to London LD4, Frankfurt and New York, Measured

Forex VPS latency from Amsterdam to the London and Frankfurt trading hubs is about 7 milliseconds round trip. In our lab, TCP connections from an Amsterdam VPS took a median of 7.36 ms to a PrimeXM server in Equinix LD4 in Slough, 7.13 ms to a server of the retail broker XTB in Frankfurt, and 73 to 75 ms to two servers near New York.

So if your broker's MT5 trade server is in London or Frankfurt, a forex VPS in Amsterdam sits 7 to 8 ms away. If it is in New York, where IC Markets keeps all its MetaTrader 5 servers, Europe adds about 74 ms and a US server is the better home. Below are the seven targets, what the milliseconds mean for slippage, and a five-minute test to run on your own broker before you order a server.

Key facts, checked on 16 October 2026:

• Median TCP connect time from our Amsterdam lab to PrimeXM's server in Slough was 7.36 ms (p95 8.57 ms, 60 samples). PeeringDB lists Equinix LD4 as the only facility of that network, AS199188.
• To Frankfurt, the medians were 7.13 ms to XTB's network and 7.27 ms to Plus.line, whose five Frankfurt buildings in PeeringDB include Equinix FR2. To New York they were 73.22 to 74.82 ms.
• IC Markets states that "All IC MetaTrader 5 servers are located in the Equinix NY4 data centre in New York" (IC Markets MT5 page).
• MetaQuotes hosts MT5 access points for brokers in Equinix London (LD4), New York (NY4) and Hong Kong (HK1), and also in Amsterdam, Sao Paulo, Mumbai, Lagos and Johannesburg (MetaQuotes announcement, 5 May 2022).
• Our own ping.pe test on 6 October 2026 measured 6.7 ms from London and 6.6 ms from Frankfurt to our Amsterdam server, and 18.6 ms from Vienna to our Prishtina server.

Forex VPS latency from Amsterdam to London, Frankfurt and New York

Forex VPS latency is the round-trip time between the server that runs your trading terminal and your broker's trade server: how long a small packet takes to get there and back, in milliseconds (ms). We timed TCP connects, the same handshake MetaTrader makes when it opens a connection, and ordinary ICMP pings, which many networks filter or answer slowly. Each run took 60 samples per target, half a second apart.

The median is the middle value of those 60 samples. The p95 is the value that 95% of samples stayed at or below, so it shows the slow moments too.

Every probe got an answer, with no packet loss. We ran the set twice, three minutes apart, from a Debian 13 container on our Amsterdam test server, and every median in the second run matched the first (shown) within 0.1 ms. The biggest change was the p95 to PrimeXM US, which rose from 77.49 to 82.00 ms.

The two London-area targets with a documented building, PrimeXM in LD4 and LINX in Telehouse North, were 0.12 ms apart, and the two Frankfurt targets 0.14 ms apart. An Amsterdam VPS is therefore about 7 ms from both cities, the same figure ping.pe measured in the opposite direction. We measured late at night with the forex market closed, when networks are quiet, so repeat the test during your own trading hours.

How a forex VPS lowers latency compared with a home PC

A VPS (virtual private server) is a rented server in a data centre that runs your terminal around the clock. It lowers latency by moving the terminal, not the broker: the trip to the trade server starts a few milliseconds away instead of from your home or office line.

The gap can be large. When we checked that our test script runs on Windows, a Windows 10 desktop PC (not a server) measured a median of 68.73 ms over 10 connects to the same LD4 server that our Amsterdam test server reached in 7.36 ms. Your home figure depends on where you live and on your provider, so run the test below from both places.

The second gain is uptime. Expert Advisors (EAs, the automated strategies MetaTrader runs) only trade while the terminal is open, and a server keeps it open through nights and weekends when your own PC is off. That is what our forex VPS plans are for.

How latency affects slippage, and who needs it low

Slippage is the difference between the price you saw when your order left and the price you were filled at. Prices keep moving while the order travels and while the broker processes it, so a shorter trip gives them less time to move.

The network is only one piece of the delay, though. The broker's processing, the liquidity at that moment and the order type matter too. Lower latency removes one source of delay; it does not promise better fills or make a strategy profitable.

Stop Loss and Take Profit levels, once set, are triggered on the broker's server, so your latency does not touch them. A trailing stop is different: the MetaTrader 5 help says it is executed in the trading platform rather than on the server, so it stops moving when the terminal is closed. That is one more reason to run terminals on a server that never sleeps.

If your strategy sits in the middle rows, our forex VPS page lists the Windows plans and the latency we measured from each city. Running bots on other markets? We measured the same question for prediction markets in our guide to Polymarket bot latency from Amsterdam.

Where are MT5 broker servers located?

Most retail brokers do not say. They publish server names, and the terminal looks up the address itself. A few publish the building: IC Markets names Equinix NY4 in New York for all its MT5 servers on its MetaTrader 5 page. PrimeXM, which sells connectivity and hosting to FX brokers, names Equinix LD4 and LD7 in London, NY4, TY3 in Tokyo and SG1 in Singapore on its infrastructure page. If your broker is silent, ask its support which city hosts the trade server for your account.

Why the ping shown in MetaTrader can mislead

An access point is a relay server that accepts terminal connections and passes them on to the broker's trade server. MT5 lets brokers put access points in front of the trade server, and MetaQuotes sells them as a hosted service in LD4, NY4, HK1, Amsterdam and other cities. Its announcement says account data is decoded only at the trading servers.

An old MT5 release note (build 346) mentions "ping to access points during their full rescan". In practice, the ping in the corner of MetaTrader may be the distance to an access point, perhaps in Amsterdam, while your order travels on to a trade server in London or New York.

How we picked the test servers

We found no broker that publishes its MT5 server addresses, and addresses from forums or old server files carry no proof of where they are or whether they still work. So we timed servers whose building is documented. RIPE Atlas anchors are small servers that the RIPE NCC places in networks worldwide as permanent measurement targets (anchor list). PeeringDB is the public database of which networks sit in which data centre, and we matched the two. PrimeXM is a fair stand-in for broker infrastructure because it hosts FX brokers. We found no public target documented in Equinix FR2 alone, so the Frankfurt rows give the city, not the building.

What adds latency between a VPS and the broker

Physics sets the floor. Light in fibre covers about 200 km per millisecond, so each 100 km of cable adds roughly 1 ms to a round trip (Lightwave). London is 358 km from Amsterdam in a straight line, a floor of about 3.6 ms; we measured 7.2 to 7.4 ms, because cables bend and every router adds a little. New York is 5,863 km away, a floor of about 59 ms, and we measured 73 ms.

We traced the routes with mtr, a tool that combines ping and traceroute, 20 probes per hop:

# as root on the Amsterdam server (Debian 13)
apt-get install -y mtr-tiny
mtr -4 -rwzbc 20 81.2.190.14

What we saw: the Frankfurt route left our network at Nikhef in Amsterdam on GSL's backbone (AS7578), passed Equinix AM5 at 0.9 ms and reached Equinix FR5 in Frankfurt at 6.6 ms; XTB's server answered at 7.0 ms on average. The route to Secaucus went via London, then across the Atlantic to Equinix NY2 at 72.0 ms, and the anchor answered at 73.4 ms. Toward PrimeXM in LD4, the first hops ran through Cloudflare's network (AS13335), and the server answered at 7.0 ms best, 8.0 ms average.

Which carriers a provider buys transit from shapes these paths. We compared two large ones in our Hurricane Electric vs Cogent comparison for Balkan and European networks.

Why we trust TCP over ping

The Sky UK row shows the difference: ICMP ping averaged 15.12 ms and never dropped below 14.97 ms, while TCP connections to the same address took 8.06 ms. Routers often give ping low priority or a different path, and a broker's server may not answer ping at all. MetaTrader talks to the broker over TCP, so TCP connect time is the closer match to what your terminal sees.

How to test your MT5 broker server ping in five minutes

Do this on the server you plan to trade from. It needs no login except to your own trading account.

1. Install MetaTrader 5, log in with the account you will trade (demo and live accounts can use different servers) and wait until the connection indicator in the status bar shows a ping in milliseconds.
2. Click that indicator and rescan the servers, so the terminal pings the broker's access points again and picks the fastest. Write down the figure. We did not open a terminal for this article, so menu names in your build may differ a little.
3. Find the address the terminal is connected to:

# in PowerShell on the Windows VPS, as the user running MetaTrader 5 (terminal open and logged in)
Get-NetTCPConnection -State Established -OwningProcess (Get-Process -Name terminal64).Id | Select-Object RemoteAddress, RemotePort -Unique

You get a small table with RemoteAddress and RemotePort columns. terminal64 is the program name of 64-bit MetaTrader 5 (terminal64.exe). We ran this command on Windows 10 against a browser process instead, because no terminal was running, and it listed each remote address and port as shown. The terminal may also connect to MetaQuotes services such as news, so if you see several addresses, test each one.

4. Save this script as tcpping.py. It opens 60 TCP connections, half a second apart, and prints the minimum, median and p95. It needs only Python 3 (python.org on Windows).

# save as tcpping.py; run as a normal user (py on Windows, python3 on Linux)
import socket, statistics, sys, time
host, port = sys.argv[1], int(sys.argv[2])
count = int(sys.argv[3]) if len(sys.argv) > 3 else 60
ip = socket.gethostbyname(host)
rtts, fails = [], 0
for _ in range(count):
    s = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
    s.settimeout(2)
    t0 = time.perf_counter()
    try:
        s.connect((ip, port))
        rtts.append((time.perf_counter() - t0) * 1000)
    except OSError:
        fails += 1
    finally:
        s.close()
    time.sleep(0.5)
if not rtts:
    print(f"{host} {ip}:{port} no answer ({fails} failed)"); sys.exit(1)
rtts.sort()
p95 = rtts[max(0, int(round(0.95 * len(rtts))) - 1)]
print(f"{host} {ip}:{port} n={len(rtts)} fail={fails} min={rtts[0]:.2f} median={statistics.median(rtts):.2f} p95={p95:.2f} ms")

5. Run it against the address and port from step 3. This example uses the PrimeXM LD4 anchor from our table, which is a public measurement target:

# in PowerShell on the Windows VPS, in the folder where you saved tcpping.py
py tcpping.py 185.97.160.7 443 60

What our Amsterdam test server printed (Debian 13, same script, 60 samples): 185.97.160.7 185.97.160.7:443 n=60 fail=0 min=6.94 median=7.36 p95=8.57 ms. On Linux, use python3 instead of py. On Windows 10 with Python 3.14 the script ran unchanged and printed the same line format.

A median under about 10 ms means you are in the right region. If MetaTrader shows 1 or 2 ms but the script shows 70 ms to another address, you are probably near an access point and far from the trade server; ask your broker's support which city hosts the trade server for your account. Only time addresses your own terminal uses or public targets like these anchors, and keep it to a few dozen samples.

Which plan and city fits your broker

All our plans run on KVM with NVMe storage, unmetered traffic and their own IPv4 and IPv6. MetaTrader needs Windows, which you choose at the configure step on VPS Mini and every VDS plan; VPS Nano and Micro run Linux only. VPS plans share their vCPUs with other servers on the host. VDS plans have dedicated vCPUs, which keeps an EA's timing steady when neighbours are busy.

RDP (Remote Desktop) is how you see and control the Windows server from your own PC. In the same ping.pe test, Frankfurt measured 27.8 ms and London 40.2 ms to our Prishtina server, so for a London or Frankfurt broker Amsterdam is the closer choice. We have no published ping.pe figures for our Dublin servers, so we leave Dublin out rather than guess.

Availability by city is on the plans page, and you can upgrade later from the client area with a short reboot. If you are unsure whether shared or dedicated vCPUs suit you, our VPS vs VDS vs dedicated server comparison explains the difference. Trading for a prop firm? Read our overview of prop firm VPS rules for 2026 before you set up.

How much RAM and CPU MetaTrader 5 needs

We did not measure a MetaTrader 5 terminal for this article, and MetaQuotes publishes no fixed figure. Its installation page says the platform runs on Windows 10/11 and that "hardware requirements depend on specific platform use conditions", such as MQL5 programs and the number of active instruments and charts.

Windows comes first: Microsoft lists 2 GB of memory as the minimum for Windows Server with the Desktop Experience, and recommends 4 GB for Windows Server 2025. So we start Windows traders on VPS Mini with 4 GB and move anyone with more than a couple of terminals to VDS Small with 8 GB. After a week of trading, check memory in Task Manager and upgrade if it sits near the top.

For the Windows side, see our Windows VPS plans and the guide to securing RDP and keeping MetaTrader running after a disconnect or restart.

Frequently asked questions

What is a good forex VPS latency?

Under about 10 ms to your broker's trade server means the VPS is in the right region. From an Amsterdam VPS we measured medians of 7.1 to 8.1 ms to servers in London, Slough and Frankfurt. Getting down to one or two milliseconds takes a server in the same data centre campus as the broker, and below one millisecond takes co-location.

What is the ping from Amsterdam to Equinix LD4?

In our lab, TCP connections from an Amsterdam VPS to a PrimeXM server in Equinix LD4, Slough, had a median of 7.36 ms and a p95 of 8.57 ms over 60 samples, and ICMP ping averaged 7.54 ms. For comparison, physics alone sets a floor of about 3.6 ms for the 358 km straight line from Amsterdam to London.

How do I check my MT5 broker server ping?

MetaTrader 5 shows a ping next to the connection indicator in its status bar, and a server rescan from that indicator's menu tests the broker's access points again. That figure can be the distance to an access point rather than the trade server, so also time TCP connections to the address the terminal uses, for example with a short Python script.

Does lower latency reduce slippage?

It removes one cause: a shorter network trip gives the price less time to move while the order is in flight. Slippage also depends on the broker's processing, liquidity and the order type, so a faster route does not promise better fills. It matters most for tick-reacting EAs and news trading, and little for trades held for hours.

Is Amsterdam good for a forex VPS?

For brokers in London or Frankfurt, yes: we measured about 7 ms from Amsterdam to both cities, and ping.pe measured 6.7 and 6.6 ms in the other direction on 6 October 2026. A server inside LD4 would be faster still to an LD4 broker. For brokers in New York, any European city adds more than 70 ms.

Where are MT5 broker servers located?

Most brokers do not publish it. The FX data centres that LMAX and PrimeXM both name are Equinix LD4 in Slough near London, NY4 in Secaucus near New York, TY3 in Tokyo and SG1 in Singapore, and IC Markets says all its MT5 servers are in NY4. MetaQuotes also runs MT5 access points for brokers in several cities, including Amsterdam. Ask your broker which city hosts your account's trade server.

Measure the broker, then pick the city

If your broker trades from London or Frankfurt, a server in Amsterdam sits about 7 ms away, and the plan matters more than the city: VPS Mini for one or two terminals, VDS Small with dedicated vCPUs once you run several. If your broker trades from New York, choose a US server, even though we do not sell one.

Run the five-minute check before you commit, then compare the Windows plans on our forex VPS page, or send us your broker's server name on Telegram and we will help you compare our cities.

← All articles