Skip to guide
NoahAI
NOAH’S GUIDE · LEARN BY DOING

How autopilot and learning connect

Understand observations, decisions, HOLD, recorded outcomes and reassessment, including how to assess improvement.

Reviewed 5 Oct 2026 · Confirm availability in your installed client.

Autopilot stays within your chosen scope

The core engine observes markets, selects candidates and checks regime, costs, profitability policies, account risk and order constraints. HOLD is a valid outcome. A custom strategy is optional.

01Observe

Read permitted market context.

02Select

Evaluate a candidate.

03Check

Apply costs, risk and permissions.

04Execute / HOLD

Use virtual PAPER or permitted LIVE paths.

05Record

Keep decision and outcome evidence.

06Review

Reassess conditions and versions.

More evidence is not guaranteed improvement

Records can reveal which regimes and conditions worked or failed. Recording data, completing training, validating improvement and applying it to LIVE are distinct steps. Provider conversations do not automatically retrain the trading engine.

How records feed back into decisions

Account/venue learning records retain observations and outcomes. Adaptive strategy paths supply regime and recent performance context to candidate-specific declared rules; they do not overwrite global TP/SL settings across unrelated strategies.

Performance-based optimization can produce review proposals for TP/SL or leverage from applicable outcome evidence. Proposal generation is separate from application and requires review, approval and revalidation. Check which paths are enabled in your installed client and venue.

If a rule repeatedly loses in ranging markets, review its regime, costs and drawdown, create a revised version and compare it in PAPER. Do not infer improvement by selecting only favorable periods.

Compare on a consistent basis

Review net PnL with recorded costs, drawdown, sample size, unseen periods and PAPER forward tests. Keep venue, version, mode and currency separate. Unknown costs are not zero.

A repeatable routine

  1. Before starting: check connections, mode, permitted targets and loss limits.
  2. During operation: review fresh data, HOLD/block reasons and protection of open positions.
  3. Afterward: review costs, outcomes and regime-specific evidence.
  4. Change one condition at a time, save a new version and revalidate in PAPER.

When profitability validation blocks entries

Read the precise reasons and sample period. In supported versions use Maintenance → Trade record checks/recovery → Profitability diagnosis → Reassess current evidence. Reconcile missing outcomes first. Do not erase loss evidence or bypass hard limits; diagnosis does not resume live orders automatically.