A fair question before you put anything near a funded account. The short answer: by design, an outage on our side can't stop your bot or move your account.
We read account state and answer "can this trade pass the rules?". This lives beside your bot, never in front of its orders. If AlgoProven is unreachable, your bot never even notices — it sends orders through its own broker path as always. The only thing you lose during an outage is the live dashboard view and new audit records (which resume on reconnect).
Guard mode can send a halt / flatten instruction to your own bot or authorized execution bridge when a rule is about to break. It never places orders itself. Because the action runs through your infrastructure, you decide what happens when our signal doesn't arrive:
Bot trades on its own local risk rules — same as no AlgoProven.
Bot drops to a conservative size / tighter local cap until we're back.
Bot flattens & pauses on its own — the safest default for a strict account.
The fallback is evaluated on your machine, so it works even if our API is completely unreachable.
The synchronous check() call your bot can make before an order is built with a timeout + fail-open default: on any timeout or error it returns "no objection" and your bot proceeds on its own local rules. A rule check should only ever tighten risk — never add a new way to fail.
audit() path is fire-and-forget and fail-open, so it never blocks or retries into your order path. Recording resumes automatically on reconnect.Because we sit beside your order path and never hold credentials, the worst case of an AlgoProven outage is "you're back to trading without us." Try the runtime on a chart you control.
More on trust: security & architecture · status · risk disclosure · terms