Cycle Lab cockpit
Cycle Lab · Explainable model
Eight states. Every verdict needs receipts.
The Regime Engine is an inspectable research framework—not a black-box prediction.
Research statev0.4 candidate · original validation preserved
From the documentaryWatch the argument, then inspect the evidence.
State machine
Regimes are transitions, not labels.
A state changes only when multiple evidence families confirm the transition.
Trend and liquidity deteriorate together.
Forced selling and volatility overwhelm orderly price discovery.
Seller exhaustion appears, but confirmation remains incomplete.
Structure improves while the first serious retest remains decisive.
Spot demand, breadth and structure confirm one another.
Participation broadens and pullbacks remain contained.
Price strength persists while conviction and breadth weaken.
Support, liquidity and participation begin failing together.
Evidence ledger
Weights and gates remain visible.
The v0.4 candidate compares the frozen original score with a confirmation gate designed around the documented 2019 and 2023 errors.
Validation design
No future information may leak backward.
Walk-forward windows, point-in-time inputs and frozen transformations will test capitulation, bottoming, recovery, bull confirmation, distribution and deterioration separately.
Transition policy
Persistence before promotion.
The next engine revision treats a state change as a process: persistence, retest quality, sequential movement and contradiction gates must remain visible.
A candidate state must survive two consecutive eligible observations before promotion.
Recovery needs a higher low plus durable participation after leverage normalizes.
The engine does not jump across intermediate states without declared override evidence.
Negative drawdown recovery and an unconfirmed long-term slope cap expansion.
A model transition creates a research assignment, never an automatic public claim.
Walk-forward validation
Regimes under pressure
2019 recovery impulse: the recovery score crossed the expansion threshold too early.
2023 recovery: the recovery score crossed the expansion threshold too early.
Original vs recalibrated
Confirmation gate comparison
Error matrix
First-pass research audit
Original accepted8/10
Original false calls2
Gate accepted10/10
Gate false calls0
The v0.4 candidate caps expansion when drawdown recovery remains negative and the 200-day slope is not yet persistent. This is a declared research hypothesis—not an optimized production rule.
01 · Bear expansion→02 · Capitulation→03 · Bottoming→04 · Early recovery→05 · Bull confirmation→06 · Bull expansion→07 · Distribution→08 · Regime deterioration
Weight simulatorResearch interaction—does not rewrite the frozen validation.
What the chart shows
Ten trailing-only historical checkpoints, the composite score and two documented premature expansion calls.
Blue Rogues Read
The model identifies severe stress well, but momentum-heavy recoveries can cross the expansion threshold before confirmation is durable.
What it confirms
The first-pass rule set is inspectable, repeatable and sensitive errors are visible.
What it does not confirm
Predictive edge, statistically complete ground truth or a production current regime.
Evidence required next
Add persistence and retest-quality gates, then rerun without changing the historical labels.
- Source
- Blue Rogues internal Episode 004 research history
- Frequency
- Selected walk-forward checkpoints
- Coverage
- 2015–2023 · ten evaluation windows
- Transform
- regime-research-v0.3 preserved · v0.4 confirmation-gate candidate
- Observed
- 2026-08-30T20:56:40.577Z
Blue Rogues boundary
Methodology Cycle research describes evidence and conditional regimes. It does not provide individualized investment advice, a guaranteed outcome or an automatic publication trigger.