Server location and my ping from London: 4 brokers measured

[FXS London Scalper]

I measured round-trip ping from my VPS in London to the trade servers of four brokers I've used. Then I measured from my home broadband for comparison.

From the London VPS:

• Two brokers with servers in the LD4 data centre: 1 to 2 ms.
• One broker in New York: about 70 ms.
• One broker that didn't publish a location: around 12 ms, which suggests somewhere in western Europe, maybe Amsterdam or Frankfurt.

From home, the best was about 15 ms and the worst over 90 ms, with noticeable jitter in the evenings.

Does it matter for scalping? For my style, with 5 to 10 pip targets on M1 and M5, the difference between 2 ms and 15 ms makes no measurable difference in fills. The difference between 15 ms with jitter and 2 ms stable showed up in my slippage logs only around the open and data releases.

The bigger point is knowing where your broker is. If you're in Asia trading a broker in London, a VPS close to the server can matter more than which broker you choose.

How do others measure this? I used the platform's own ping display plus a simple tcp check.

[Amanhasnoname]

I did a similar comparison from a VPS in Frankfurt last year and saw almost the same pattern. Brokers in the same data centre were under a millisecond, the ones in London sat around 9 to 10 ms, and a broker that turned out to be routing through New York was nearly 90. Like you, I couldn't see any difference in fills between 1 and 10 ms for normal M1 entries.

What did show up was jitter at home. My broadband looked fine on average, but during evening streaming hours the ping would jump from 20 to over 150 for a few seconds at a time. That's exactly when a stop modification or a quick exit fails to reach the server in time, and it happened to me twice on news before I moved to the VPS for live trading.

For measuring, I used a small script that opens a TCP connection to the trade server port every few seconds and logs the time it takes, then I looked at the 99th percentile rather than the average. The platform's own ping figure is useful, but it updates slowly and hides the spikes.