Measure the clock
The terminal reports calendar times in the broker server’s timezone, which may shift with daylight saving. Releases that have long been fixed at a New York wall-clock time, such as payrolls at 08:30 and ISM at 10:00, make good anchors: compare where the terminal puts them with where they should be, for every season of every year, and you have measured the offset instead of assuming it. Record the result and its confidence, and refuse to trust a measurement with too few anchors.
Use the schedule and not the outcome
A calendar API returns the actual value as it stands today, after every revision, so a surprise feature built from history is contaminated, most of all for the releases revised most. The forecast field can also be empty for older rows. Release times have none of those problems: they are published weeks ahead, never revised, and complete as far back as the calendar goes. Using tomorrow’s release time on today’s bar is not a leak.
A blackout rule in one line
Stay out for five minutes before a release and add nothing new until fifteen minutes after. Filter by importance, and scope by the currencies of the chart symbol. Because the bar’s opening time is the resolution, use blackout windows on M1 to M15. In the News module that rule is AWNewsBlackout(5, 15).
When the calendar runs out
If the calendar stops covering the future, a naive all-clear rule concludes the coast is permanently clear. A safer rule reports “not clear” whenever it cannot be evaluated, so a stale file stops trading instead of permitting it. Keep the live file current with a service that rebuilds it on a schedule and replaces it only when its own checks pass.
This is a timing tool. It does not promise to keep any account inside a loss limit. The funded-account worked path lists what it does not model.