Trade Copier Latency: How Fast Does Copying Need to Be?

The copier is rarely the slow part — the follower broker's own execution usually is. Where the delay actually comes from, and when it changes a trade.

7 Aug 2026 · 5 min read

Fast enough that the follower order is on its way the instant the source fill is reported — and after that, the copier is usually not the slow part. The follower broker's own execution is normally the binding constraint, and no copier, cloud or local, can shorten it. This guide splits the delay into the legs that make it up, and sets out when the gap genuinely changes a trade.

Where the time actually goes

A copied trade passes through three stages. They are not remotely equal in size, and only one of them is really a property of the copier.

• Source fill detection — how the tool learns the source position opened or closed. A copier subscribed to a real-time stream hears the fill as it happens; one polling on a timer can sit on a full poll interval of dead time before it even starts. This is the stage where copier design genuinely differs.
• Dispatch — scaling the lot, applying any symbol remap and allowlist check, and sending the order to each follower. The arithmetic is trivial; the network hop to the broker gateway is the real cost, and it is small.
• The follower broker's execution — the order arrives, queues, is priced against that broker's liquidity, and may be requoted or rejected. Outside any copier's control, and typically both the largest leg and the most variable.

Put plainly: a copier competes on the first stage, barely on the second, and not at all on the third. Any comparison that ignores the third leg is comparing the small parts.

Why "milliseconds" in the marketing is mostly meaningless

Published latency figures rarely say what they measured. Whether the number covers only the tool's internal processing or the whole path to the broker, whether it is a single run or a distribution, and what the market was doing at the time are usually all left unstated. None can include the follower broker, because that number belongs to a different company and changes with the session and the instrument. So a copier advertising a headline figure and one advertising nothing may put your order at the same broker at the same moment.

If a vendor quotes a number, ask what it measures and what it excludes. If the answer is vague, treat the figure as decoration. We deliberately don't publish one for Trepeat: a number with the broker leg stripped out would flatter us and tell you nothing useful about your own accounts.

A different fill price usually isn't lag

The most common misreading is a source filling at 1.24010 and a follower at 1.24022, and concluding the copier is slow. Most of the time that gap is spread and liquidity, not timing: two brokers quoting the same instrument at the same instant print different prices, and each fills against its own book.

A rough test — if the difference is consistent, in the same direction, and roughly the size of the spread gap between the two brokers, that's pricing. If it's large, erratic, and only on fast moves, timing is a plausible suspect. Why follower fills differ works through the full list of causes, including requotes and news-time slippage.

When latency genuinely matters

• Scalping — positions held for seconds to a few minutes over a handful of pips. Any delay is a meaningful fraction of the whole move, so the dispatch stage earns real attention.
• News releases — spreads widen and requotes get common in the seconds around a print. Delay compounds with slippage here, though it's the broker leg that dominates, which is exactly the part no tool can fix.
• Very fast round trips — a position opened and closed inside seconds can see the close arrive while the copy is still opening. What matters is that the copier handles that rather than leaving a position behind on the follower.
• Exits in general — a late entry costs you a slightly worse price; a late exit can cost considerably more. Judge exit handling harder than entry speed.

One rules note: some brokers' terms restrict latency-based tactics such as arbitrage between venues. That's a different activity from repeating your own trades, but the line is drawn by your broker's agreement, not by us and not by this page — read it, and see is copying your own trades allowed for how to approach the question. Nothing here is legal advice.

When it doesn't matter much

If you hold positions for hours or days and enter at levels rather than ticks, a second either way changes nothing you can measure. Swing traders routinely over-index on speed and under-index on faithfulness. What actually decides whether copying works for that style is whether closes and partial closes copy as reliably as opens, whether sizing lands correctly on accounts of different sizes, and whether a symbol name mismatch quietly drops trades. Start from what a trade copier actually does if you're weighing that up.

Cloud vs VPS isn't the deciding factor

A common assumption is that a copier running on a VPS beside the broker's server must be quicker. Physical distance matters far less than the two things above — event-driven versus polled detection, and the broker's own execution, which is identical either way. A local copier still sends its order over the internet, and carries constraints of its own: terminal open, machine awake, process running. Cloud vs local (VPS) copiers compares the two on uptime, cost and upkeep, which are the differences that show up in practice.

What to check instead of a speed number

• Is fill detection event-driven, or polled on a timer? Ask directly.
• Are closes and partial closes copied, or only opens?
• Can one action close the source position and every copy of it, or do you unwind followers by hand?
• When a copy doesn't land, do you find out immediately — with the reason the broker gave — or notice it at the end of the week?
• Can you see each follower's actual fill, rather than assuming it matched?

Trepeat dispatches on real-time streams rather than a polling timer, copies opens, closes and partial closes, and raises an alert naming what the broker reported when a copy doesn't land. It connects MetaTrader demo-server accounts — the type prop firm evaluations and funded accounts run on — and refuses live accounts at connect time. What it can't do is set your broker's execution speed, and any tool claiming otherwise is describing something it doesn't own.