Trend and liquidity deteriorate together.
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 statev0.3 illustrative · historical validation in progress
State machine
Regimes are transitions, not labels.
A state changes only when multiple evidence families confirm the transition.
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 remain visible.
Sensitivity tests will vary these weights and measure false signals before any current classification can be treated as production-grade.
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.
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.
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
- 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.