Skip to main content
By asset · Forex

Trading forex: where your broker really lives

Your broker isn’t down the street from you — it’s in a data centre in Slough, Sydney, or New Jersey, and the route from your desk to that rack is what determines your fill quality.

TL;DR — the real story

IC Markets, Pepperstone, Exness, FXTM, and XM all run their primary MT4/MT5 servers at Equinix LD4 near London. Your distance to that building determines your baseline. But for most traders, raw distance isn’t the problem — jitter is. A jittery path during NFP or a London open costs you more than a steady 80 ms connection ever would.

Typical broker hub London (Equinix LD4)
Trader-sensitive moments Session opens, NFP, FOMC
Biggest real enemy Jitter, not latency
Section 1

Where forex brokers actually host

The dominant answer for retail forex is London. Equinix LD4 in Slough — about 20 miles west of central London — is where most of the brokers you’ve actually heard of keep their MT4 and MT5 infrastructure. IC Markets, Pepperstone, XM, FXTM, Exness, FxPro, and Tickmill all run primary execution servers there. LD4 and its adjacent campus LD5 collectively hold more retail forex server capacity than any other single location on earth.

There are two other significant hubs. For Australian-facing accounts, IC Markets and FP Markets run nodes at Equinix SY4 in Sydney — if you’re in Melbourne or Singapore and your account is with the ASIC entity, you’re probably talking to Sydney. For US-facing brokers — OANDA US, TD Ameritrade FX, and Interactive Brokers — the hub is Equinix NY4 in Secaucus, New Jersey, the same campus that connects directly to NASDAQ and NYSE.

Prop firms are the notable exception. FTMO, The Funded Trader, Topstep, and MyForexFunds all run their MT challenge environments in London — but on separate server infrastructure from their retail broker counterparts. If you’re prop trading, you’re almost certainly routing to LD4 as well.

You don’t have to guess. Open MT4 or MT5 and look at the server hostname in your login window — something like ICMarketsSC-Live04 or Pepperstone-Live03. Ping that exact hostname from your terminal for 60 seconds. The average round-trip time and the min–max spread tell you your actual situation, not a theoretical one.

  • IC Markets: Equinix LD4 (London) for global; SY4 (Sydney) for AU/APAC accounts
  • Pepperstone: Equinix LD4 (London) primary; LD5 secondary
  • Exness / FXTM / XM: Equinix LD4 (London)
  • OANDA (US): Equinix NY4 (Secaucus, NJ)
  • FP Markets: Equinix SY4 (Sydney) for APAC clients
  • FTMO / prop firms: Equinix LD4 (London) on separate challenge server pools
Section 2

Why forex is specifically latency-sensitive

Forex isn’t automatically more latency-sensitive than every other asset — but the conditions that make latency matter are baked into how forex works.

The first condition is tight spreads. On EURUSD, the world’s most liquid pair, raw spreads from ECN brokers routinely sit at 0.0–0.2 pips. When your spread is that narrow, slippage of even 0.3 pips from a slow fill represents a cost that scales quickly. A scalper taking 20 trades per day on EURUSD with consistent 0.3-pip slippage is paying 6 pips per day purely to network delay — before they’ve made a single directional call.

The second condition is session concentration. Forex volume doesn’t distribute evenly across 24 hours. The London open at 08:00 UTC and the London/New York overlap from 13:00 to 17:00 UTC are when the real action happens for EURUSD, GBPUSD, and USDJPY. During these windows, broker systems are under maximum load, spreads are tightest, and prices reprice the fastest. Your connection is also competing with every other trader hitting the same broker servers at the same time.

The third condition is scheduled news events. NFP on the first Friday of each month, CPI releases, and FOMC meetings compress enormous price moves into 1–5 seconds. Your order needs to reach the broker’s matching engine before the price has moved past your entry. If your connection adds an extra 200 ms of jitter at that exact moment — which residential ISPs commonly do under load — your fill is either requoted or arrived too late.

The scalper math

M1 and M5 scalpers on tight pairs are disproportionately affected because their entire trade thesis can expire in under a second. If you’re trading M15 or above and giving the market room to move, a 100 ms variance in your connection is unlikely to be what’s hurting you. If you’re scalping news, it almost certainly is.

Section 3

What the real bottleneck actually is

Most traders assume the problem is distance — that because they live far from London, they’re at a disadvantage. That’s partly true, but it misses the bigger issue. Distance is fixed. Jitter is variable. And it’s the variable part that causes the fills you can’t predict.

Here’s how a typical residential internet path to Equinix LD4 behaves: on a quiet afternoon, your round-trip time might be 45 ms. Clean, steady, well within usable range. During London open — when traffic peaks on both your ISP’s network and the broker’s systems — that same path can start swinging between 35 ms and 200 ms on the same connection. That swing is jitter. The average looks fine. The fills don’t.

The reasons are structural. Residential ISPs route through shared transit networks where the path to London changes based on load, peering agreements, and what every other customer on your segment is doing. You might be sharing transit capacity with a content delivery network, a gaming network, and a business VPN — none of which care about your EURUSD order timing.

WiFi makes it worse. A 2.4 GHz connection in a building with six other networks adds 5–30 ms of local jitter before your packets even reach your router. For discretionary traders, Ethernet is free and eliminates an entire category of variance.

What to measure

Don’t ping Google. Ping your broker’s actual MT server hostname for 60 seconds at three different times: your local morning, London open, and NY open. Compare the min–max spread at each time. If the spread is wider at the busy windows, your ISP’s routing is your problem, not your distance.

Section 4

What actually helps — in honest order

1. VPS at your broker’s data centre (for EA traders)

If you run EAs — grid bots, scalping systems, news traders — a VPS at Equinix LD4 or a nearby London facility is the right answer. Your MT4 or MT5 terminal sits metres from the broker’s matching engine. Round-trip time drops to 1–3 ms. Your EA fires at consistent intervals. The path is stable because it’s a dedicated link inside a data centre, not a shared residential route. Cost is $10–25/month depending on provider. The trade-off is that you’re trading via remote desktop — your charts and interface live on a remote machine.

2. Proxy for discretionary traders

For manual traders, the VPS approach trades one friction (bad routing) for another (remote desktop). A proxy is better: you keep your own machine, your own chart layout, your own keyboard shortcuts — but your trading traffic exits through a node that’s physically close to your broker’s data centre and uses a dedicated, uncongested path. The round-trip time won’t be 1 ms, but it will be consistent. Consistent is what helps fills. TradersProxy has nodes at LD4, SY4, and NY4 — matching the three major forex broker hubs.

3. Ethernet over WiFi

Free and underestimated. Plug in. You eliminate a layer of local jitter before your packets even leave your building. If you’re already using Ethernet and still seeing variance, the issue is further upstream in your ISP’s network.

4. Check your broker’s server hostname

Open MT4 or MT5. Login screen — look at the server field. That hostname is where your orders actually go. Run ping ICMarketsSC-Live04 (or your equivalent) for 60 seconds. If you see large swings, you have a routing problem. If you see a steady 1–5 ms, you’re already on a VPS near the server. If you see a steady 40–60 ms with minimal variance, your connection is usable as-is.

How TradersProxy fits

For discretionary forex traders who want to keep their own setup but stop losing fills to ISP routing noise — this is the middle option. No remote desktop. No new machine. Just a cleaner path.

Your traffic exits through a node matched to your broker’s data centre — LD4 for most retail brokers, SY4 for APAC, NY4 for US. See plans or read the argument first.

Questions traders ask

Forex execution FAQ

  • For most retail forex brokers, the answer is London. IC Markets, Pepperstone, Exness, FXTM, XM, and FxPro all run their primary MT4/MT5 infrastructure at Equinix LD4 in Slough, near London. IC Markets also has a Sydney node at Equinix SY4 for APAC clients. OANDA and some US-facing brokers run at NY4 in Secaucus, New Jersey. Open MT4 or MT5, look at your server hostname in the login window, and ping that hostname for 60 seconds to confirm exactly where you're talking to.

  • It depends on how you trade. For swing traders holding positions for hours or days, 80 ms versus 30 ms is irrelevant. For scalpers working M1 and M5 charts on EURUSD, GBPUSD, or USDJPY during session overlaps, jitter is directly visible in slippage. The bigger the news event — NFP, CPI, FOMC — the more a slow or inconsistent connection costs you, because spreads widen and repricing happens in milliseconds.

  • Jitter is the variation in your round-trip time. If your broker server is always 45 ms away, that's stable and predictable. If it swings between 20 ms and 180 ms depending on the moment, that's jitter — and jitter causes inconsistent order fills, unexpected slippage, and EA behaviour that doesn't match your backtest. Most residential ISPs produce more jitter than a dedicated proxy or VPS because they route through shared transit networks that vary under load.

  • A VPS at Equinix LD4 is the best pure-latency solution if you run automated EAs — your MT4/MT5 terminal sits physically close to your broker's server, so round-trip time drops to 1–3 ms. But you trade via remote desktop, which adds friction. A proxy is better for manual traders: you keep your own machine, charts, and keyboard layout, and only the trading traffic takes a clean route. Most retail forex traders who trade discretionarily get more practical value from a proxy.

  • The London/New York overlap — roughly 13:00 to 17:00 UTC — is the highest-volume, most latency-sensitive window. EURUSD, GBPUSD, and USDJPY are most active during this period. The London open (08:00 UTC) is also critical for GBP pairs. Major data releases — US NFP on the first Friday of each month, CPI, and FOMC meetings — compress slippage into a matter of milliseconds. If your connection is going to fail you, it will fail you at these moments.

  • Open MT4 or MT5 and look at the server field in your login window. You'll see something like ICMarketsSC-Live04, Pepperstone-Live03, or Exness-Real14. Ping that hostname from your terminal or command prompt for 60 seconds. If the average is under 10 ms, you're probably close to the datacenter or already using a VPS. If it's 80–200 ms with variance, you're far away or experiencing routing problems. A traceroute shows every hop the packets take.

  • Yes, more than most traders expect. WiFi introduces variable latency — your 2.4 GHz signal can add 5–30 ms of jitter from interference alone, separate from your ISP's routing. During a news event, that extra variance on top of your already variable ISP path means your orders arrive at inconsistent times. If you trade M1 or M5 during news, Ethernet is a cheap, effective fix — free if you already have a cable.