Most MT5 broker servers live in three buildings worldwide: Equinix LD4 in London, NY4 in New Jersey, and TY3 in Tokyo. Here is why, and how to test your latency to them.
Every order you send travels: your computer → ISP → internet backbone → broker's server → broker's routing layer → LP gateway → LP matching engine, then the response travels the same path in reverse.
The first half of that journey (your computer to broker server) is your responsibility. The second half is the broker's. The latency adds.
For typical retail trading on M5/M15 charts with holding times measured in hours, an extra 100ms each way does not matter. For:
...latency starts to matter measurably. 200ms latency means you see a price 200ms after it became valid, and your reaction reaches the server 200ms after you act. The market can move several pips in that window.
The hub for European FX. Located in Slough, just west of London. Houses servers for:
If your broker is CySEC, FCA, or BaFin-regulated, their MT5 server is most likely in LD4. The proximity to LP gateways means broker-to-LP latency is sub-millisecond.
The hub for North American FX. Located in Secaucus, just across the Hudson from Manhattan. Houses:
Some brokers with global customer bases keep their MT5 server in NY4 specifically to serve US-region traders well. If you are in North America and trading with an EU broker, latency to LD4 is 70-100ms from the US East Coast.
The hub for Asian FX. Located in Otemachi, central Tokyo. Houses:
If you are trading from Singapore, Hong Kong, or Sydney, latency to TY3 is typically 50-80ms. To LD4 from the same locations, latency is 200-280ms.
Some brokers host outside the big three:
"Which data centre is my MT5 server located in?" Most institutional brokers will answer directly. Retail brokers may be vague ("we have multiple servers globally"), in which case ask again about the specific server you are logged into.
Some brokers encode location in the server name: Broker-Live-London, Broker-NY-01, BrokerName-TYO-Demo. Not always, but worth checking.
If your broker tells you the server hostname (e.g. live01.broker.com), you can resolve it to an IP and trace route. Last hops often show the city or data centre.
# Windows
tracert live01.broker.com
# Mac / Linux
traceroute live01.broker.comLook at the last 2-3 hops. ISP names like "Equinix" or "Telehouse" indicate data centre. City names in DNS rDNS (like edge-router.lon01.example.net) indicate London.
Look at the bottom-right of the MT5 window. There is a green icon labelled with milliseconds (e.g. 16/4 ms). The first number is round-trip ping to the broker server. The second is some MT5 internal metric (often unused).
Watch this number through a trading session. If it varies wildly (16ms one minute, 200ms the next), your local network is congested. If it is consistently high, your geographic distance is the issue.
For more precise measurement, ping the broker server hostname from a command line:
ping live01.broker.comThis gives you the network round-trip without MT5's internal overhead.
| Range | Quality | Typical setup |
|---|---|---|
| 0-5 ms | Excellent | VPS in same data centre as broker |
| 5-30 ms | Very good | VPS in same metro region, or home connection in same city |
| 30-100 ms | Good | Same continent, residential broadband |
| 100-200 ms | Acceptable for non-time-sensitive trading | Different continent, residential broadband |
| 200-500 ms | Poor | Wifi congestion, far geographic distance, or both |
| 500ms+ | Unworkable | Severely degraded connection. Diagnose and fix. |
If you are running EAs that need low latency, the standard solution is a VPS (Virtual Private Server) in the same data centre as the broker. Major options:
For full comparison see our MT5 VPS comparison.
If you are physically located in a different region from your broker, you face an unfixable baseline latency. From Singapore to LD4 is approximately 180ms one way regardless of network quality. A VPS in Singapore gives you fast local access but does not solve the broker-server distance.
Solutions:
Geographic distance is not the only latency factor. Your ISP's peering relationships with major backbone providers determine actual routing.
Example: an ISP in Sydney might route Sydney to London via a transcontinental cable through Los Angeles, adding 50ms over the direct (theoretical) path. A competitor ISP might route via direct Sydney-Singapore-London cables, saving the 50ms.
If your latency is significantly worse than expected for your geography, your ISP's peering is suboptimal. Switching ISP (or using a VPN to a better-peered location) can help.
Only if your strategy requires fast execution. For swing trading on H4 or D1, location is largely irrelevant. For M1 scalping or news trading, it matters.
In MT5, View > Servers shows the connection list. The IP is not displayed directly but you can find it by resolving the hostname via your operating system's DNS tools.
Yes, slightly. A broker co-located with their LPs gets the freshest prices and can pass them through faster. A broker with cloud-based servers far from LPs sees stale prices and adjusts spreads wider for safety.
A VPN does not reduce latency. It typically adds latency due to the extra hop. VPNs solve a different problem (geo-blocking, privacy) and are not a latency solution.