CFDs are complex instruments and come with a high risk of losing money rapidly due to leverage. Trade only with money you can afford to lose.
Open Exness Account →
Measured data

Where the Wait Starts: The Part of Execution That Never Leaves Your Desk — Lesotho

Everything below the click happens before an order reaches anyone else: the window has to respond, the processor has to be free enough to notice, and the local connection has to carry it out of the building.

Open Exness Account →

100+ instruments  ·  Founded 2008

A click is not the beginning of an order, it is the middle of one. Before anything leaves the machine the window has to respond to the press, the program has to be given processor time to act on it, and the local connection has to carry it outward. None of that stretch belongs to the venue at the other end, none of it appears in a figure measured there, and it is the only part of the wait a desk can shorten by itself.

What the figures below do and do not cover

The table underneath describes the stretch that begins once an instruction has left the machine it was typed on. It says nothing about the stretch before that, which is entirely local: the window has to redraw, the processor has to be free enough to register the click, and the local connection has to accept the outbound request. That earlier stretch belongs to the desk, is invisible to any measurement taken at the other end, and is the only part a workspace can actually change.

Execution speed and fill quality vary with market conditions, liquidity and position size.

Measured round-trips by instrument and size

InstrumentLot sizeAvg executionSlowest fillAvg slippage (signed)Better / zero / worse fillsRejects
EUR/USD0.01151 ms187 ms-0.3 pts1 / 2 / 00
EUR/USD0.1151 ms172 ms-0.3 pts1 / 2 / 00
EUR/USD1146 ms172 ms-0.3 pts1 / 2 / 00
GBP/USD0.01141 ms156 ms+0.7 pts0 / 1 / 20
GBP/USD0.1130 ms140 ms+0.0 pts0 / 3 / 00
GBP/USD1135 ms156 ms+0.3 pts0 / 2 / 10
XAU/USD (Gold)0.01141 ms172 ms+0.0 pts0 / 3 / 00
XAU/USD (Gold)0.1151 ms172 ms+80.0 pts1 / 1 / 10
XAU/USD (Gold)1130 ms141 ms-72.3 pts2 / 1 / 00

Across 27 measured round-trips the average fill took 130–151 ms depending on instrument and size, with 0 rejected orders in total. Negative slippage means the fill was better than the quoted price at the moment the order was sent.

Slippage in points, signed: negative = filled better than quoted, positive = worse. ‘Rejects’ counts orders the platform refused at that size.

How this was measured

  • Real market orders (buy, then immediate close) placed in an Exness MetaTrader 5 terminal.
  • Latency timed in-terminal from order send to broker confirmation.
  • Sizes stepped 0.01 → 1.00 lot to expose size-dependent slippage.
  • Fills depend on market liquidity, account and conditions, and can differ.

Measured in-terminal on Exness’s own MetaTrader 5 pricing feed and symbol specifications, refreshed on a schedule. All figures are indicative and change with market conditions.

Open Exness Account →

Three things happen on your side before anything is sent

The first is drawing. A window that is busy redrawing twenty charts is a window that responds to a press a moment later than an idle one, and on a machine near its limit that moment is not small. It is also entirely self-inflicted, which is the good news.

The second is scheduling. A program only acts when the processor gives it a turn, and it shares those turns with everything else running — a browser with thirty tabs, an update installing itself, a video playing in the corner. The third is the local link: the first stretch of a connection out of a building is usually the slowest and the most variable stretch of it.

Why this part is invisible to any figure taken elsewhere

A measurement taken at the receiving end begins when something arrives there. Anything that happened before arrival is outside the window being measured, so a desk with a slow local stretch and a desk with a fast one produce identical figures at the far end while feeling completely different to use.

This is worth knowing mainly so that the two are not confused. A number describing the far end is a fact about the far end; the experience of pressing a button is a fact about the room you are sitting in, and improving one does not move the other.

What a workspace can actually change

The levers are unglamorous and they are all local. Fewer open charts and fewer indicators leave the window free to respond. Closing whatever else is competing for the processor does the same. A wired connection instead of a shared wireless one removes the most variable stretch of the local path.

None of this is a claim about outcomes and none of it is worth chasing beyond the point where the desk feels responsive. The reason to do it is narrower: it is the portion of the wait that is yours, so it is the portion that responds to anything you do.

Frequently asked questions

Which part of the wait is mine and which part is not?
Everything up to the moment the instruction leaves the machine is local: window redraw, processor availability and the first stretch of the connection. Everything after that is elsewhere.
Does a busy workspace really change how a click feels?
Yes, because the same window that is redrawing charts is the window that has to register the press. A machine near its limit answers a moment later, and closing charts is the direct test of it.
Why does a figure measured elsewhere not describe my experience?
Because it starts counting when something arrives at the other end. The local stretch happens before that and sits outside the measurement entirely.
Is a wired connection worth the trouble?
It removes the most variable part of the local path, which is the point rather than raw speed. Variability is what makes a desk feel unpredictable.
How do I know when the local side is as good as it needs to be?
When the window answers a press without a visible pause and closing more charts stops making a difference. Past that point the remaining wait is not on your desk.

Related Exness pages