How the engine works
A plain-English explanation of the Regime Intelligence framework — what it measures, how it classifies market conditions, and what it cannot do.
What Regime Intelligence does
Markets don't move randomly. They move in regimes — extended periods where the same forces dominate: volatility, credit stress, momentum, and cross-asset correlation. Regime Intelligence detects which regime is active right now, across every asset class, at every timeframe, updated every hour.
This is not a trading signal. It is a market context layer — a live read on whether conditions favor risk-taking or risk reduction, whether stress is building or easing, whether the current move is noise or a regime shift.
Market Regime Explained
Each timeframe is assigned exactly one market regime. The engine ranks today's Composite Stress Score (CSS) against that rolling history and expresses the result as a 0–100th percentile stress rank. CSS blends volatility, drawdown, momentum stress, and related inputs into one score — it measures market stress, not price direction.
The rank uses a rolling history of past CSS values on that same timeframe and asset. Window length is calibrated per timeframe and asset class — roughly three years of history at the long horizon (Wl): about 756 daily bars for US equities and large-cap stocks, about 1,095 daily bars for crypto on 1D, and proportionally longer windows on intraday bars.
Every asset at every timeframe is classified into one of these five states. Labels reflect stress relative to that asset's own history (CSS percentile bands), not whether price is rising or falling.
Important: A regime describes the current market environment — not whether price is bullish or bearish. An asset can rally or decline in any regime. The label reflects the level of market stress relative to that asset's own history, not the expected direction of the next move.
Regimes are relative to each asset's own historical behavior. Clear Skies for Bitcoin does not imply the same level of absolute volatility as Clear Skies for Treasury bonds or large-cap equities.
To reduce unnecessary regime changes, all labels require persistence before transitioning. Higher-stress regimes generally require stronger confirmation before clearing.
| Regime | Stress rank | Meaning |
|---|---|---|
| 🌤 Clear Skies | Lowest ~30% | Quiet conditions |
| 🌬 Tailwind | 30–50% | Orderly market |
| 🧊 Thin Ice | 50–70% | Stress building |
| ⛈ Storm Warning | 70–90% | Elevated stress |
| 🌪 Full Storm | Top ~10% | Extreme stress |
In practice: Volatility is unusually quiet for this asset on this timeframe. Price action is generally smoother, daily swings tend to be smaller, and volatility shocks occur less frequently. Strong trends can still develop, but the market environment is not being dominated by elevated stress.
Engine rule: CSS ranks below approximately the 30th percentile — current stress is lower than roughly 70% of recent observations for this asset and timeframe.
In practice: Volatility is present but remains generally supportive of orderly price discovery. Trends often extend without the instability seen in higher-stress environments, while pullbacks are less likely to become disorderly. Market conditions remain constructive, although direction should always be confirmed using other indicators.
Engine rule: CSS ranks between approximately the 30th and 50th percentiles — lower-to-middle historical stress for this asset on this timeframe.
In practice: The ground is beginning to crack. Volatility is increasing, market conditions become less stable, and reversals become more common. Breakouts may become less reliable as uncertainty builds, requiring greater selectivity and risk awareness.
Engine rule: CSS ranks between approximately the 50th and 70th percentiles — upper-middle historical stress. Early-warning zone where stress is becoming meaningful but has not yet reached extreme levels.
In practice: Market stress has become elevated. Price swings widen, gap risk increases, and technical behavior becomes less predictable. Traditional chart patterns and technical setups often become less dependable until stress begins to subside.
Engine rule: CSS ranks between approximately the 70th and 90th percentiles — within the highest 20% of historical stress for this asset and timeframe.
In practice: The market is experiencing exceptionally high stress. Large price swings, heightened gap risk, and sharp increases in volatility can dominate price behavior. Correlations may shift unexpectedly, and many traditional technical reference points become less reliable while stress remains elevated.
Engine rule: CSS ranks above the 90th percentile (not including 90 itself), or an immediate crash override is triggered by severe drawdown or volatility-spike rules within the engine.
Full tutorial with live examples: Market Regime Explained.
What drives the classification
The engine processes six independent channels of market information, each capturing a different dimension of stress:
Realized and implied volatility across equity (VIX), bond (MOVE), and cross-asset markets. Measures how uncertain markets are about future prices.
High-yield and investment-grade credit spreads (OAS). When companies must pay significantly more to borrow, financial conditions are tightening — a leading signal of economic stress.
Price momentum across assets. Measures whether markets are trending in a clear direction or breaking down.
Peak-to-trough losses across the asset universe. Identifies whether stress is isolated or systemic.
Extreme move detection. Captures crash signals (sudden sharp declines) and melt-up alerts (parabolic rises that often precede reversals).
When assets that normally move independently start moving together, it signals systemic stress. The 2008 crisis was characterized by everything falling simultaneously.
The Composite Regime Score is a weighted combination of all six channels, calibrated against five decades of historical crisis periods.
Multi-timeframe analysis
Regime Intelligence classifies every asset across five timeframes simultaneously: 15-minute, 1-hour, 4-hour, daily, and weekly.
This matters because different forces operate at different speeds. A daily regime can be in Full Storm while the 15-minute regime shows Tailwind — meaning short-term buying pressure within a longer-term crisis. Or the weekly regime can be Clear Skies while the daily drops to Thin Ice — an early warning of deterioration before it shows up in the longer view.
The Timeframe Agreement Score (TAS) measures how aligned the five timeframes are. When all five agree, the signal is high conviction. When they diverge, the situation is transitional — act with caution.
The Multi-Timeframe Signal synthesizes all five readings into a single directional call: Bullish, Cautiously Bullish, Neutral, Cautiously Bearish, or Bearish.
Cross-asset intelligence
No asset exists in isolation. The regime engine monitors 5566 assets simultaneously — US and international equities (including S&P large-cap names), government bonds, credit, gold, oil, industrial metals, currencies, and cryptocurrencies — and uses their collective behavior to classify each individual asset.
When gold, bonds, yen, and Swiss franc all rally simultaneously, that is a flight-to-safety signal that context-adjusts the regime reading for every equity and credit asset in the universe. When high-yield spreads widen at the same time VIX spikes and momentum breaks, the weight of evidence drives a regime downgrade even if the price chart alone looks recoverable.
This cross-asset context is what separates Regime Intelligence from a simple momentum or volatility indicator. Single-asset signals produce too many false positives. The regime framework requires multiple independent channels to confirm before classifying a period as stress or crisis.
Update cadence and data sources
The regime engine runs every hour, 24 hours a day, 7 days a week. Each run ingests fresh price data, recomputes all six channels, recalculates the Composite Stress Score, and reclassifies every asset at every timeframe.
Equities, ETFs, FX, commodities, crypto
Rates, spreads, inflation, economic indicators
Derived indicators including yield curve spreads, real yields, and proprietary FX stress indices are computed by the engine itself.
The Regime Persistence Score estimates how long the current regime is likely to last based on historical state durations, using a Cox proportional hazard model calibrated to prior cycles.
What this is not
Regime Intelligence does not predict prices. It does not tell you when to buy or sell specific securities. It does not provide investment advice of any kind.
What it provides is context — an evidence-based read on market conditions that helps you ask better questions: Is this a good environment to add risk? Is the stress I'm seeing in my portfolio idiosyncratic or systemic? Is this a dip to buy or the beginning of something worse?
The decisions remain yours. Regime Intelligence just ensures you are making them with more information than the crowd has.
Not financial advice · Regime classifications are for informational purposes only · Full disclaimer