Forex VPS latency explained: what the "1 ms" claims leave out, and a five-minute test you can run

We sell a forex VPS. There is no latency figure to any broker on this page, deliberately.

What the number is

When the terminal sends an order, the request goes to the broker's trade server, the server processes it, and a confirmation comes back. The time for the whole loop is latency, in milliseconds. MetaTrader shows it at the bottom right, next to the connection bars.

Three things decide it. Distance comes first: light in fibre covers roughly 200 km a millisecond, and Delhi to Singapore is about 4,000 km of cable, so there is a floor of around 20 ms each way that no provider can get under. Routing comes second: packets rarely take the shortest line, and a provider peered directly onto the national backbone and the international cables gives a shorter, steadier path than one that bounces through three transit networks. Then the broker's server has to do its work, and at a news release that can add more than the network does.

Why "1 ms" is nearly always marketing

A one-millisecond round trip means the VPS is in the same building as the broker's matching engine, usually an Equinix site such as LD4 in London or NY4 in New York. A few specialist providers really do rent space there and really do see sub-millisecond figures to the brokers in that building. For every other broker, and from every other location, the physics does not allow it.

Yet the comparison pages list 1 ms against nearly every provider. The fine print reads "to many major brokers" or "with some brokers", and no test is published. When you see an unqualified 1 ms, treat it as a reason to test harder.

Low against steady

A chart of latency across a trading day tells you more than the best reading. A VPS sitting at a steady 40 ms with no spikes gives predictable fills. A home connection that reads 25 ms at midnight and 300 ms when the household starts streaming gives slippage at exactly the wrong moment. Datacenter links, DDoS scrubbing that absorbs an attack instead of passing it through, and a provider that does not oversell its ports all go into steadiness. None of it appears in a headline figure.

The test

1. In MetaTrader, Ctrl+O, Server tab, note the server name.
2. On the candidate VPS, with a test IP or trial, install MetaTrader, log in to a demo on that server, read the ping in the status bar.
3. Take it four times: the Asian session, the London open at 1:30 p.m. IST, a scheduled news release, and the New York open at 7 p.m. IST.
4. Run ping and tracert from the VPS to the broker's hostname as well, so you can see the route and where the time is spent.
5. Compare providers on the four readings and the spikes.

Where a Noida server comes out

From Delhi NCR a Noida desktop, or from Mumbai a Mumbai one, is a few milliseconds away, which is why the terminal feels like it is on your desk. To the broker, our network peers directly onto the national backbone and the international cables out of India, so the route is short and steady for brokers with Asian or India-facing servers. A strategy that depends on sitting next to a New York matching engine needs a New York VPS, and we will say so when asked.

Questions we get asked

Does latency matter for a swing trader? Hardly. What matters is that the EA is running when the stop is hit.

Is lower always better? Lower and steady. Lower with spikes is worse than slightly higher and flat.

Will a VPS stop requotes? The ones caused by a slow or unstable link on your side. Not the ones caused by the broker's pricing.

Why does my ping move during the day? Internet routes change with load, and the broker's server gets busier. Hence four readings.

Can I have a test IP? Yes, from support, before you buy.