Discretionary is fine until your click is slower than a headline. I have been exploring lightweight auto-flat: calendar API or local injector that flattens / blocks new orders inside a blackout window. Not a full EA — a seatbelt.
Questions for people who have built or bought this:
Which calendar feeds have you found reliable enough not to false-flat you all week?
Do you flatten all symbols or map events → pairs (NFP → USD & gold only)?
How do you handle revised times / early releases without trusting a single source?
Prop-account angle: does auto-flat help compliance or create “EA policy” headaches?
I still want human judgement for re-entry after the print; I only want the machine to stop me being a hero at T−0. Pointers to robust setups welcome — no black-box sales pitches please. If you run something home-built, how do you test it without discovering the bug on a live NFP minute?
LondonScalper wrote: Do you flatten all symbols or map events
If you're on MT5, the built-in economic calendar is worth trying before any external API. CalendarValueHistory and CalendarEventByCurrency return events by currency with an importance level, and the times come in trade server time, so mapping to symbols is a simple currency check rather than a separate list. Two catches. The calendar functions don't work in the strategy tester, so testing the auto-flat needs past events exported to a file and read from there. And it's MetaQuotes data, one source, so a second feed for comparison still makes sense on tier-one days, which also covers the revised-times question. On mapping, events to currencies is the sensible default, with gold added for anything tier-one in USD. Flattening everything on every red event quietly costs good trades on pairs that had nothing to do with the release.
I built something like this for myself, and the biggest lesson was to map events to currencies instead of flattening everything. Flattening all symbols on every red event meant the tool closed a gold trade for a Canadian number, and I switched it off within a week, which defeats the point.
My mapping is simple: each event has a currency, each symbol has two, and only symbols containing that currency are touched. Gold and indices get a manual list, so US data hits gold and US indices, and that's it.
For times, I store everything in UTC and re pull the calendar at 07:00 and 12:00 London. If two sources disagree on a time, the tool uses the earlier one, which occasionally flattens me a few minutes early but has never left me in.
On the prop side, I'd ask the firm in writing before using it. A flatten only utility is usually fine, but the word EA in a support ticket gets very different answers depending on who reads it.
For testing, every Sunday I create a fake event three minutes ahead on a demo and watch what closes. Have you decided whether it blocks new orders too, or only flattens?
Fairman, the MT5 built-in calendar angle is the one I should have tried before wiring an external feed. CalendarValueHistory and CalendarEventByCurrency already tag importance and land in server time, so the currency-to-symbol map is almost free.
The tester catch is real. I export the prior week's events to a CSV on Friday and the auto-flat reads that file in the tester, because the live calendar functions are empty there.
I still keep a second feed on tier-one days exactly for the revised-time case you flagged. MetaQuotes alone has been fine on quiet weeks and late once on a CPI reprint window.
LondonScalper, on your block-new-orders question I ended up doing both: flatten open risk and refuse new tickets inside the blackout. Flatten alone left a forgotten buy stop that filled on the spike.
I am still unsure whether to trust the importance flag as published. On labour or inflation theme weeks I override yellow events for my pairs rather than wait for the calendar to paint them red.
The built-in MT5 calendar is a good starting point, Fairman. A detail that helps with testing: export the past events to a file once, and have the EA read that file in the strategy tester while using the live calendar in real trading. It keeps the logic identical in both cases. It is also worth checking that the event times are adjusted correctly around DST changes, since a one-hour error on an auto-flat rule could leave positions open into a release.