Method:
- The Journal tab logs a line when a request is sent and another when the deal executes, both with millisecond timestamps.
- Export a month of Journal logs (they live in the terminal's Logs folder as daily text files) and pair request and execution lines by order number.
- Group the delays by London hour and compute the median and the slowest 5%.
What to expect, broadly: quiet hours fastest, the London open and the US data window slower, and news minutes with a long tail that can reach hundreds of milliseconds or more.
Why it matters: on ordinary minutes, tens of milliseconds change nothing for a 5-pip target. In the first 30 seconds after 08:00 London, a slow tail coincides with fast prices and wide spreads, which is a good argument for waiting a couple of minutes after the open.
If someone has a cleaner way to measure it than the Journal, I'd like to hear it.