Candidate Bots And Controls
Learn what a candidate bot represents and which controls belong to the customer rather than the research engine.
The candidate bot is the bridge between research output and customer operation. It packages research into an object the operator can review, compare, and act on without exposing raw internal machinery.
What a candidate bot carries
A candidate bot is not just a raw run result. It is a product-facing object that includes:
- The strategy family or display identity
- The current evidence posture
- The suggested next step
- The range of controls the customer is allowed to use
Customer controls
The operator chooses between Backtest, Paper, and Live. They can also Pause and Resume at any time.
- Market scope: selected symbols or all eligible symbols
- Directionality where supported
- Sleeve size inside approved ranges
Account selection when multiple exchange accounts are connected to the workspace.
What stays institutional
The customer should not be asked to control internal research state, promotion ladder logic, feature engineering internals, validation semantics, or data-source mechanics. Those belong to the research system.
Reading the card
When a candidate bot shows a warning or suggestion, the operator should understand it as:
- A statement about evidence and launch posture
- Not a hidden statement about changing the internal research ladder itself
The workspace lets the user decide where to run, how to run, how much exposure to take, and when to pause. It does not ask the user to rewrite the research engine.