Between your click on Buy and the broker's server lie thousands of kilometers of glass, dozens of routers and one hard law of physics. We break the packet's journey down piece by piece: where every millisecond comes from, where your traffic actually travels, how to measure ping honestly — and the only way to reduce it that actually works.
When you hit Buy, you start a race: the order has to reach the broker's server, and the confirmation has to make it back. All that time the price keeps living its own life. How long the trip takes is decided not by your tariff or the “power” of your internet, but by geography: thousands of kilometers of fiber, dozens of routers, and the speed of light — which can't be bought or overclocked. Let's walk the packet's road inch by inch, from your router to a rack in a data center.
Ping (also RTT, round-trip time) is how long a small packet takes to reach a server and come back. It's measured in milliseconds and has almost nothing to do with the “internet speed” on your plan: bandwidth is the width of the pipe, ping is its length. A gigabit plan downloads a file faster, but it doesn't move you a single millimeter closer to a server in New York.
Ping is a sum of segments, and the sum is distributed very unevenly. Wi-Fi and your home router contribute 1–10 ms, the city and backbone paths a few more — and then geography takes over: every ocean costs tens of milliseconds, and no tariff cancels that.
A packet doesn't travel in a straight line. The internet is thousands of autonomous systems (the same ASNs from our proxy article) that negotiate traffic exchange: sometimes directly, sometimes through exchange points (IX), sometimes via transit operators. The route is chosen by the BGP protocol — based on agreements and cost, not geometry. So a packet from Moscow to New York almost certainly detours through Frankfurt or London, and when a backbone fails it suddenly goes “the long way around” with an extra 30 ms.
Across oceans, traffic rides submarine cables: there are about five hundred of them, carrying nearly all of the planet's intercontinental traffic. On the main financial route, London — New York, milliseconds cost hundreds of millions: the Hibernia Express cable was laid in 2015 along a deliberately straightened arc — around $300 million to beat neighboring cables by ~5 ms. The buyers of that head start are banks and HFT funds.
Average ping is like average temperature across a hospital. The real picture is the spread (jitter) and packet loss. A lost packet gets resent by TCP, and instead of 100 ms the order travels 300–500: in the terminal it looks like a random freeze. A network that holds a steady 25 ms is better for trading than one that shows 15 but spikes to two hundred once a minute.
The broker's website almost certainly lives on a CDN in another country. The real trading server's address is visible in the terminal — that's the one to measure.
mtr or WinMTR show every hop: where the route detours, at which node latency grows and packets get lost.
A ping at 3 a.m. tells you nothing. Measure during the European and American sessions and on news releases — when everyone's network is loaded.
A few hundred samples at minimum. Spread beyond ±10 ms or loss above 0.5% is worth investigating even if the average looks pretty.
The same measurement from a server in a data center near your broker shows how many milliseconds you leave on the road from home every day. In OnlyForex this is built-in mechanics: the Connections section lists every server of your broker with its country and a live ping — and you should measure from the VPS itself, because your home channel plays no part in the battle route.
Ping doesn't get “optimized” — it shrinks together with distance. While the terminal sits at home, you carry the whole route with you: the last mile, the ISP, the ocean. Moving the terminal to a VPS in a data center next to the broker's server removes the route almost entirely — what's left is the next rack over and fractions of a millisecond. Home becomes a remote control: the RDP connection to the VPS can cross an ocean for all it matters — it has no effect on order execution speed.
For arbitrage, placement is an even finer question. The engine lives between two points — the quote feed node and the broker's server — and the ideal VPS sits in the middle of that path, so both delays fit into single-digit milliseconds. OnlyForex feed nodes run in London, New York and Tokyo; the best case is when all three points land in the same financial hub like NY4 or LD4. And a terminal at home, in this coordinate system, is a voluntary 50–100 ms head start handed to your counterparty on every order.
Where to rent the server, why the location is chosen for a specific broker, and how the Connections section finds all of its servers with live pings — covered in a separate article: “A remote server for arbitrage: picking a VPS without overpaying”.
Prime time: the neighbors are streaming, your ISP's network and your last mile are loaded. Packets wait in router queues — both latency and jitter grow. One more reason to measure ping during the hours you actually trade.
No. The plan is bandwidth — the width of the pipe. Ping is determined by route length and queues along the way. The pipe gets wider, not shorter: New York is still thousands of kilometers away.
For trading — cable. Wi-Fi adds one to ten milliseconds, but the worse part is jitter spikes under interference — and stability matters more to a terminal than the average.
It depends only on distance: fractions of a millisecond within one data center, 1–5 ms across a city, tens across a continent, over a hundred across an ocean. Check your measurement against the table above: if it's notably worse than the norm for your distance, the problem is in the route — find it with a traceroute.
Game servers sit in your region — hundreds of kilometers away. Trading servers sit where the liquidity is: New York, London, Tokyo. That's thousands of kilometers, and no setting cancels that physics.
It almost always increases it: an extra intermediary on the route, as we covered in the proxy article. The rare exception is when your ISP's route is so crooked that a VPN accidentally straightens it — but don't build your trading on that; the reliable fix is a VPS next to the broker, not a tunnel.
Ping can't be bought with a tariff — only shortened with distance. How to pick a VPS in the right data center, check pings to broker servers and assemble the whole stack up to the first trade — it's all in the guide; the first chapters are open.