The execution speed number on the broker website is probably not your number

[Shadow Trader]

"Average execution speed: 30 milliseconds." It's on nearly every broker landing page, and it's close to meaningless on its own.

Where was it measured? Probably from inside the broker's own data center to their own server, with no internet path involved. Which orders were included? Often everything, including small-size orders from clients sitting next to the servers. During what period? Usually a quiet stretch.

What matters for a scalper is the round trip from click to confirmation, under the conditions that actually count: during news, at open, with real size, from a real home connection.

That number can be five or ten times worse than the advertised one. And the average hides the tail. A fill that takes 40 ms ninety-five times and 1.5 seconds five times looks fine in the average and ruins the five trades where it counts.

The honest approach is to measure it directly. Timestamp the click, timestamp the fill, and log it. Percentiles tell more than averages: the median, the 95th, and the worst.

There's an additional wrinkle: some brokers measure from the moment the order enters their trading server until it is sent to the liquidity provider, which leaves out the entire bridge and the provider's response time. A claim of 20 ms may describe only the first inch of a long road.

Another trick is the choice of sample. A busy broker might report the average for all orders, including thousands of tiny market-maker-hedged ones that never leave the building. The scalper's orders, going outside, take longer.

A decent measurement routine uses a script that sends a small order, records the local time, receives the fill confirmation, and logs the difference. Run it at six different times of day for a week, and the picture becomes quite different from the brochure.

Is anyone actually logging their own execution latency, and what's the worst number seen?