autonomous-research-loop

You can automate an autonomous research workflow by orchestrating bounded, paper-only research loops that maintain clear separation and explicit labeling between baseline and experiment paths to generate data-driven reports.

Safety constraints for running long-running research loops include enforcing paper-first defaults, keeping live trading disabled, and requiring explicit experiment labeling to ensure strict separation between baseline and experiment paths.

Yes, the autonomous research loop supports EXE packaging, startup flow, and operator-panel actions on Windows to reproduce research states and observations within the Meridian repository.

To start a bounded research loop in paper-only mode, use the operator panel or startup script to initiate the workflow, which provides operator visibility through compact summaries and strict safety defaults.

Use a bounded research loop instead of live trading automation when you need to safely compare baseline and experiment paths, as this workflow enforces disabled live trading and paper-first defaults to prevent unintended execution.