Latency in retail trading is usually quoted as one number: the round trip time between your VPS and the broker's trade server. It is easy to measure, easy to advertise, and it is only one segment of the path an order actually takes. The full sequence is your expert advisor deciding, the platform building the order, the network carrying it, the broker's server receiving and validating it, the dealing logic accepting or rejecting it, and a confirmation coming back. Network time is often the smallest part of that.
This matters because traders optimise the number they can see. Moving from 40ms to 4ms feels like a tenfold improvement and changes nothing if the broker's own processing takes 80ms and its price feed updates every 250ms. It changes a great deal if you are trading a strategy whose edge lives inside the first second of a news release, where a queue position measured in milliseconds decides whether you get the price on screen.
Where the milliseconds actually go
Physical distance sets a floor. Light in fibre covers roughly 200 kilometres per millisecond, and real routes are not straight, so London to New York realistically costs something in the region of 35 to 45ms round trip no matter what you pay. No VPS provider can beat physics. What they can do is remove the bad routing, the congested consumer link and the machine that sleeps.
The second component is the broker's side. Order validation, margin checks, risk desk logic and, on some accounts, a deliberate delay before confirmation. This is invisible from a ping test and only measurable from your own execution records. If your fills consistently arrive hundreds of milliseconds after the order, the network is not your problem.
The third is your own code. An expert advisor that recalculates indicators over 5,000 bars on every tick, or that opens a file, or that runs a loop over all open orders before every decision, can burn more time than the entire network path. Profile it before you buy hardware.
Match the location to the server, not to the marketing
The only location that matters is the one closest to your broker's trade server, and that is frequently not where the broker's head office or licence is. Many brokers host in the same handful of financial data centres, with London, New York and Tokyo carrying most retail forex. Ask support directly which data centre your account's trade server sits in, or read the server address and check where it resolves.
Then pick a VPS in that facility or in a provider with a direct route to it. A machine in the same building typically shows a round trip of a couple of milliseconds. A machine on the right continent shows single digits to low tens. A machine on the wrong continent shows the physics penalty, and no amount of CPU fixes it. The general setup work is covered in the forex VPS guide.
If you run several accounts at brokers hosted in different regions, one VPS cannot be optimal for all of them. Either accept the compromise or run a second machine near the second server. Running a copier across the gap adds the latency of both legs, which is a point traders miss in copier setups.
Which strategies actually care
Be honest about this before spending. A swing strategy holding positions for days does not care whether an order takes 5ms or 90ms, and buying a premium VPS for it is spending money on the wrong problem. What that trader needs is uptime, so the platform is running when the alert fires, and a stable machine that survives a reboot.
Latency starts to matter for intraday systems trading around fast moves, and it dominates for scalping and news strategies where the price on screen exists for a fraction of a second. Those are also the strategies most exposed to slippage and requotes, and a faster machine changes only the part of that caused by your own delay. If a broker widens spreads around a release, arriving 20ms earlier gets you a wider spread sooner.
How to test before you commit
Most providers offer a trial or a monthly term. Use it and measure four things rather than trusting a marketing page.
Measure the round trip to the trade server from inside the VPS, at your actual trading hours, over several days rather than once. Measure it again during a scheduled news release, since congestion shows up under load and not at 6am on a Sunday. Record your platform's own reported execution time on real orders on a small account, because that includes the broker's processing. Finally, watch CPU and memory while your expert advisors run, since a machine sold on latency and starved of resources will queue your own code before the network ever gets involved.
Write the results down and compare providers on the same instrument at the same hours. Two machines advertised identically frequently differ by an order of magnitude on the only route you care about.
The things that break more often than latency
In practice, expert advisors fail for dull reasons far more often than for slow networks. The VPS reboots for updates overnight and the platform does not restart. The platform restarts but the chart template does not load and the advisor is not attached. A password change on the account leaves the terminal disconnected. Disk fills with logs. The provider migrates the machine and the IP changes, breaking an allowlist.
Fix those first: automatic platform startup with the correct profile, a monitor that alerts you when the terminal is not connected, log rotation, and a written note of every setting that would need to be re-entered after a rebuild. A machine at 30ms that runs continuously beats a machine at 2ms that quietly stopped on Tuesday. The operational side of running automated systems is covered in expert advisors explained.
What to spend
Buy the cheapest machine in the right data centre with enough memory for the number of terminals you actually run, and pay monthly until you have measured it for a full month including a news week. Providers discount annual terms heavily, which is exactly why the annual commitment should come after the measurement rather than before. Automated trading carries the same high risk as manual trading, and faster execution changes the cost of entry, not the quality of the strategy behind it.
"People buy a VPS in London because the marketing said London. Half of them are trading with a broker whose server is in New York, and they have made their latency worse by paying for it."
— Alex Onta, Executive Director, SINGUARD
Key Takeaways
- Ping to the trade server is one segment of the path; broker processing and your own code often cost more.
- Choose the VPS location by where your broker's trade server is hosted, not by the broker's head office.
- Swing strategies need uptime rather than milliseconds; scalping and news systems are the ones that feel latency.
- Test round trip and real execution times across several days including a news release before committing to an annual term.
Frequently Asked Questions
What is a good VPS latency for trading?
A machine in the same data centre as the broker's trade server typically shows a couple of milliseconds, and one on the same continent shows single digits to low tens. Cross-continent routes are limited by distance and cannot be improved by paying more.
Does a VPS improve my fills?
It removes delay caused by your own connection and by a home machine sleeping or losing power. It does nothing about the broker's processing time, spread widening or market liquidity, which are usually the larger part of a poor fill.
Do I need a VPS if I trade manually?
Rarely for latency. A VPS is worth having if you run expert advisors, copiers or alert-driven tools that must stay online continuously, since the value there is uptime rather than speed.
About the Author
Alex Onta is an Executive Director at SINGUARD. He built eTrader, the terminal, the mobile apps, eTrader Broker, Copytrading, Business and Community, along with the worldwide clustered-server infrastructure it all runs on, with his brother Roman Onta helping on the design, and he leads that division today. Together with Roman he builds the Prop Firm CRM, the Broker CRM, Scalegram and CopySignals, and the two of them carry worldwide compliance, payment processing and international business structuring side by side. He lives and works in Dubai for most of the year. Meet the executive duo leading Singuard's five divisions.