spec.md 1.7 KB
ADDED Requirements
Requirement: Blocked glued weak rebounds SHALL be reviewable as a concrete dataset
The workspace SHALL provide a repeatable way to replay the strategy and collect every buy_block_glued_high_weak_rebound candidate with subtype and forward observation metrics.
Scenario: Blocked-candidate replay
- WHEN the review script replays the official strategy on the current history
- THEN each blocked glued weak rebound is recorded with:
- signal date
- subtype
- signal indicators
- regime snapshot
- short-window followthrough metrics
Requirement: Followthrough probe lanes SHALL be analyzable in isolation
The workspace SHALL provide subtype-isolated probe outputs for delayed glued followthrough reentries.
Scenario: Probe-lane comparison
- WHEN the review script runs subtype-isolated probe configs
- THEN it publishes reentry-trade outputs for each lane independently
- AND lanes do not need to be mixed to produce conclusions
Requirement: Timing and exit-treatment evidence SHALL be published before strategy changes
The review SHALL compare timing sensitivity and exit-side aftermath for followthrough reentries.
Scenario: Reentry trade review
- WHEN a followthrough probe trade exists
- THEN the review records:
- same-close return
- next-open return
- post-entry max-close opportunity
- post-exit followthrough left on the table
Requirement: RC1 default behavior SHALL remain unchanged
This change SHALL NOT alter RC1 default strategy outputs.
Scenario: Research-only delivery
- WHEN the change is applied
- THEN no default RC1 branch or governance path is modified
- AND the change is limited to research/reporting artifacts