What is a Latency Test?

Q: What is a Latency Test?

A latency test measures the round-trip time from your terminal to the broker's server, in milliseconds. Here's how to run one and why it matters.

A latency test measures how long an order takes to travel from your terminal to the broker's trade server and back, in milliseconds. It is a round trip: request out, acknowledgement in. The lower the number, the sooner your order reaches the market and the closer your fill lands to the price you saw. For most manual traders it is a background detail; for scalpers, news traders and high-frequency EAs it is the difference between the price you clicked and the price you got.

What you are actually measuring

Latency is the delay on the wire, not the speed of your strategy. It depends mostly on the physical distance and network path between your terminal and the broker's server — which is why where your VPS sits matters far more than how fast your home internet is. A 2ms round trip from a data centre next to the broker beats a 90ms round trip from a fast home line every time.

Ping and latency get used interchangeably. Ping is the tool; latency is the number it returns.

How to run one

In MetaTrader. The connection status in the bottom-right corner shows a ping figure to your current server. MT5 lists ping per server in the login dialog, so you can compare before you commit.
From the VPS. Most forex VPS providers publish the latency from their data centre to the major brokers, and many run a live test page. Check the figure to your broker's server, not a generic one.
Broker tools. Some brokers show the trade server's location or offer a ping utility. Match your hosting to that location.

Run the test at the times you actually trade. Latency during a quiet Asian session and latency across a London or New York open are different numbers.

A single reading tells you little. Test several times across a normal trading day and watch the worst case, not the average — the spike during high volatility is when slippage hurts most.

What good looks like

Under 5ms — your terminal is in the same data centre as the broker. Ideal for scalping and news EAs.
10 to 30ms — same region. Fine for most automated strategies.
80ms and up — a different continent to the server. Workable for swing systems, poor for anything time-sensitive.

A worked example

Your EA scalps for 5 pips a trade and takes 400 trades a month. On a 90ms connection it averages 0.6 pips of slippage against you on entry. Move it to a VPS beside the broker at 3ms, and that falls to about 0.1 pips. The half-a-pip saving, at roughly £0.80 a pip on 0.10 lots, is £0.40 a trade — around £160 a month the old setup was quietly handing back. The test that revealed it took a minute.

The Karnek note

Karnek reads your account read-only and sits outside the order path, so it adds no latency of its own. It shows you the fills and slippage your setup actually produced, so you can judge whether a latency test is worth acting on.

See it in Karnek: Karnek's health monitoring tells a dead VPS apart from a stalled EA.