How to Use Regime Research
A plain-language guide to structural market monitoring — what each state means, how the detection system works, and what historical transitions have looked like.
Part 1: Regime Pulse & Regime Scan — Structural Detection
- The core idea
- The four directional states
- Detection phases and conviction
- What historical transitions looked like
- False positives and limitations
- Sector ETFs
- How investors have historically used this kind of information
Part 2: Regime Lens — Quantitative Bitcoin Valuation
This section covers the S&P 500 structural health monitor (Regime Pulse) and the sector ETF monitor (Regime Scan), both built on the critical transitions detection framework.
1. The core idea
Complex systems — ecosystems, power grids, financial markets — share a common property: they show measurable changes in their statistical behavior before major transitions.
Think of it like a bridge. A healthy bridge absorbs vibrations quickly. A structurally compromised bridge keeps vibrating longer after each truck passes, and the vibrations get larger. You can measure this change before the bridge fails.
Financial markets exhibit the same phenomenon. Before major drawdowns, the statistical “recovery rate” of market returns tends to slow down, and fluctuations tend to grow. These changes are measurable. They’re not predictions of where the market will go — they’re structural measurements of how the market is currently behaving.
This phenomenon is part of critical transitions theory, documented across dozens of domains in peer-reviewed research since 2009. Regime Research applies it to financial markets.
2. The four directional states
Every trading day, the engine classifies the S&P 500’s structural condition into one of four states:
No structural deterioration detected. The warning signs are absent. CLEAR does not mean the market will go up. It means the structural warning signs monitored by the system are not present. The system spends roughly a quarter of its time in CLEAR, with the remainder split across the active monitoring states.
Statistical signatures of structural weakening have appeared. The system is losing resilience — its ability to absorb shocks is declining. This is an early signal. Many DETERIORATING periods resolve without a drawdown.
The statistical signatures that have historically preceded major drawdowns are present. This is the system’s highest-severity classification. Not every ALERT leads to a drawdown — false positives are inherent to early warning systems — but historically, most major market declines were preceded by this state.
The ALERT conditions have subsided and structural health is stabilizing. The system uses a calibrated recovery window before returning to CLEAR. Historically, RECOVERING periods have shown strong returns as markets normalize after stress.
Important: These states describe structural conditions, not market direction. CLEAR does not mean “buy” and ALERT does not mean “sell.” The market can decline from CLEAR and rise during ALERT. These are measurements of structural health, not trading signals.
3. Detection phases and conviction
The engine monitors two independent channels of structural health, each tracking a different statistical dimension. When these channels fire matters:
QUIET
Neither channel is elevated. This is the baseline condition during CLEAR periods.
EARLY DETECTION
One channel — the one monitoring distributional characteristics — has identified stress, but the other channel hasn’t confirmed. This can occur while the market appears calm on the surface. Many early detections resolve without escalation.
CONFIRMED REGIME
The primary structural channel has identified deterioration independently. This produces a standard ALERT classification. Across all ALERT episodes, 84% have been followed by a decline of 5% or greater.
ESCALATION
The strongest configuration: one channel identified distributional stress first, and the primary channel independently confirmed structural breakdown within 30 days. When this sequential pattern has occurred historically, the hit rate has been substantially higher than standalone signals. This is the system’s highest-conviction detection.
The detection phase tells you how the current state was reached, not just what the state is. Two ALERT periods can have very different conviction levels depending on the underlying detection pattern.
4. What historical transitions looked like
Applied to the S&P 500, the engine has identified structural deterioration preceding 10 of 11 major drawdowns since 1997. The median lead time — the gap between the first ALERT signal and the onset of the drawdown — was 47 calendar days.
Some examples of what the system detected and when:
Before the 2007–2009 financial crisis, the system entered ALERT in October 2007, approximately 126 days before the most severe phase of the decline. Before the COVID crash of 2020, the system entered ALERT approximately 14 days before the drawdown. Before the 2022 rate-shock selloff, structural deterioration appeared months ahead of the peak.
The one miss: September 11, 2001 — a sudden external attack with no statistical precursors in any domain. This type of event is outside the system’s detection capability by design.
Hindsight bias warning: All historical detection data is backtested. These results were developed by applying the current methodology to historical data — they were not generated in real time. Backtested results are inherently optimistic. The live system may perform differently. Detailed backtesting methodology and limitations are documented on the dashboard.
5. False positives and limitations
The system produces false positives. Periods where structural deterioration is detected but no significant market decline follows. This is not a flaw — it is an inherent property of early warning systems. A smoke detector that never false-alarmed would also miss real fires.
Across all 43 ALERT episodes since 1997, 36 (84%) were followed by a decline of 5% or greater. The remaining 7 episodes (16%) resolved without a significant drawdown. Users should expect false positives and plan accordingly.
Other limitations to be aware of:
The system is backward-looking
It measures what has already happened to the market’s statistical structure. It cannot anticipate exogenous shocks (pandemics, geopolitical events, policy surprises) that haven’t yet manifested in price data.
Regime states can persist
States can persist for extended periods. The system is in a non-CLEAR state approximately 74% of the time — this is an active monitoring system, not a passive one that occasionally fires. The system provides the most value around transitions. If the dashboard shows the same state for weeks, the system is working correctly — it is continuously measuring and confirming the current structural condition.
It is not a timing tool
Detecting structural deterioration weeks or months before a drawdown doesn’t tell you what happens in that interval. Markets can rally significantly after an ALERT begins. The system identifies when structural conditions have changed, not when prices will change.
6. Sector ETFs
Regime Scan applies the same structural detection framework to individual sector ETFs, but with an important difference: each sector has been calibrated independently based on how the signals perform in that specific market.
Not every sector responds to structural detection the same way. Some sectors produce too many false positives with the standard framework, while others have specific detection patterns that work better than the general approach. The three sectors currently monitored — Energy, Utilities, and Financials — each use different alert filtering rules based on what the data validated.
Sectors with high correlation to the S&P 500 (Technology, Industrials, Healthcare, and others) are deferred because their signals are largely redundant with the broad market monitor. The sectors included are those where independent, non-redundant structural information was validated.
7. How investors have historically used this kind of information
Regime detection is one input among many. Different investors with different goals have historically used structural health information in different ways. None of the following constitutes investment advice — these are descriptions of how structural health information has been applied in practice.
As a risk overlay
Some investors use regime state as one factor in their overall risk assessment. During ALERT periods, they may reduce position sizes, tighten stop losses, or increase hedging — not because the system tells them to, but because their personal risk framework assigns more weight to structural deterioration.
As a rebalancing input
Some investors use regime transitions as a trigger to review portfolio allocation. A shift from CLEAR to DETERIORATING doesn’t necessarily prompt action, but it may prompt a review of exposures and assumptions.
As context for other signals
Regime state can provide context for other indicators. A valuation signal during a CLEAR regime has different implications than the same signal during ALERT. Some investors use structural health as a filter for their existing decision-making process.
As a calm-market discipline tool
For investors prone to anxiety during normal market volatility, a CLEAR classification can provide evidence-based reassurance. The system is specifically designed to distinguish between normal market noise and genuine structural deterioration.
Disclaimer: Regime Research does not provide investment advice. The descriptions above are educational — they describe how structural health information has been used in practice, not how you should use it. You should consult with a qualified financial advisor before making any investment decisions.
8. Regime Lens — Quantitative Bitcoin Valuation
Regime Lens applies a different analytical approach to Bitcoin. Rather than structural health monitoring based on critical transitions theory, Regime Lens uses the Monetary Power Index (MPI) — a dual-model valuation framework that combines two independent pricing methodologies into a unified fair value estimate, valuation corridor, and dual undervaluation signal.
The Monetary Power Index
The MPI combines two independent valuation models into a single framework. Each model approaches Bitcoin pricing from a different angle, and their agreement or disagreement provides additional analytical information beyond what either model offers alone.
The valuation corridor
The two models produce a valuation corridor with three reference levels: a floor (the lower bound of fair value), a fair value estimate (the central tendency), and a ceiling (the upper bound). The current price’s position within this corridor provides context for whether Bitcoin is trading above or below the models’ estimates of fair value.
The dual undervaluation signal
When both models independently agree that Bitcoin is significantly below fair value — and the oscillator confirms deep undervaluation — the system generates a dual undervaluation signal. This signal has preceded positive 12-month returns in all 24 completed episodes since 2015. Two additional episodes are currently within their 12-month evaluation window.
The signal is not a buy recommendation. It identifies periods where both independent valuation approaches agree on significant undervaluation. Past performance does not guarantee future results.
Halving cycle position
The system tracks Bitcoin’s position within the four-year halving cycle (Accumulation, Expansion, Distribution, Late Cycle), providing additional context for the valuation readings.
Structural regime overlay
Regime Lens also includes a structural regime monitor adapted from the same framework used in Regime Pulse. In Bitcoin markets, this signal has historically shown an inverse relationship — structural instability has preceded upside moves rather than drawdowns. This overlay provides additional context but is not the foundation of the BTC valuation framework.
Important: Bitcoin is a highly volatile asset. The valuation models and structural signals are analytical tools, not forecasts. Cryptocurrency markets carry substantial risk of loss. Past signal performance does not guarantee future results.