COVOL Composite Index — User Guide
Aggregate risk measurement and systematic stress decomposition across four COVOL analyses
Contents
Tool Overview
Essential Concepts
How to Use
Interactive Features
Data Interpretation
Practical Applications
Understanding Data
Troubleshooting
Tips & Best Practices
Essential Concepts
Before diving in, here are the key concepts you'll encounter:
COVOL (Common Volatility)
A statistical measure of the magnitude of common shocks to the volatility of financial assets. COVOL captures global events — such as geopolitical tensions, policy changes, and pandemics — that cause widespread, simultaneous price movements across different asset classes, sectors, and countries. The model separates each asset's volatility into a shared common component and an idiosyncratic component, estimating how sensitive each asset is to these common shocks.
Why it matters: Traditional diversification cannot mitigate COVOL risk because a single event can increase volatility for all assets in a portfolio simultaneously. Assets with low COVOL sensitivity are potentially valuable portfolio diversifiers. See the Methodology section below for the full mathematical framework.
PCA (Principal Component Analysis)
A statistical technique that finds the main pattern driving multiple measurements. Think of it as finding the 'common thread' across markets.
Why it matters: PCA extracts the shared systematic factor from individual analysis signals, enabling measurement of market-wide vs. analysis-specific risk.
PC1 (First Principal Component)
The dominant pattern found by PCA. In this context, it represents market-wide stress affecting all analyses simultaneously.
Why it matters: High PC1 means stress is shared across analyses. The PC1 loadings show exactly how much each analysis contributes to this shared stress.
CAV (COVOL-Adjusted Volatility)
COVOL-Adjusted Volatility measures the annualized risk of the common factor within each COVOL analysis, incorporating both the GARCH baseline volatility and the COVOL surprise factor. CAV values are interpretable as annualized volatility of the PC1-weighted portfolio within each analysis group — typically 10-25% in normal periods and 100-300% during crises. See the Methodology section below for the full formula.
Why it matters: CAV translates the abstract COVOL multiplier into an intuitive annualized volatility number, making risk levels directly comparable across analyses.
Loading
How strongly each analysis connects to the common factor (PC1) at a given point in time. Loadings are stored as time series, capturing how relationships evolve.
Why it matters: Shows which analyses are most connected to broad market stress and how those connections change over time.
ACAV (Aggregate COVOL-Adjusted Volatility)
Aggregate CAV combines the four individual CAV measures into a single portfolio-level risk metric using risk-parity weights and rolling correlations of GARCH-COVOL standardized innovations. It represents the overall risk level across all four COVOL analyses. See the Methodology section below for the full derivation.
Why it matters: Provides a single aggregate risk reading that accounts for both individual analysis volatilities and their correlations, making it the headline measure for overall market conditions.
Distribution Strip
A visualization showing a dot positioned on a horizontal density strip, indicating where the current value falls within its historical distribution.
Why it matters: Provides immediate visual context for whether a current reading is normal, elevated, or extreme relative to history, replacing the previous progress-bar style.
Raw COVOL
The estimated latent common volatility factor, normalized to mean = 1. Values greater than 1 indicate that common volatility shocks are larger than average; values less than 1 indicate quieter conditions.
This is the \(x_t\) in the COVOLmodel: \(e_{it} = [s_i(x_t - 1) + 1] \cdot \varepsilon_{it}\).
Why it matters: Raw values preserve tail information. A 20x event remains a 20x event, ensuring extreme common volatility shocks are not compressed by smoothing transforms.
How to Use
Methodology
The COVOL Composite Index builds on V-Lab's COVOL model to produce risk measures that are intuitive, comparable across analyses, and sensitive to both persistent volatility and sudden common shocks. This section walks through the three main building blocks: the underlying COVOL model, the individual COVOL-Adjusted Volatility (CAV) for each analysis, and the Aggregate CAV that combines them into a single portfolio-level measure.
Each equation is accompanied by variable definitions and an explanation of the intuition behind it.
The COVOL Model
At the heart of the composite index is V-Lab's COVOL model. For each of the four analysis groups, the model examines the volatility residuals of every asset in that group and asks: how much of each asset's unexpected volatility can be explained by a single common shock that hit all of them at the same time? The model separates each asset's volatility residual into a shared component (driven by the common factor) and an idiosyncratic component (unique to that asset):
where:
- \(e_{it}\) is the observed volatility residual for asset \(i\) at time \(t\). This is the part of the asset's return that the GARCH model did not predict
- \(x_t\) is the latent common volatility factor, normalized so its long-run average is 1. When \(x_t > 1\), common shocks are larger than usual; when \(x_t < 1\), conditions are calmer than average
- \(s_i \in [0,1]\) is the factor loading for asset \(i\), measuring how sensitive that asset is to the common shock. An asset with \(s_i\) close to 1 is highly exposed; one with \(s_i\) near 0 is largely unaffected
- \(\varepsilon_{it}\) is the idiosyncratic shock, the random component unique to each asset after accounting for the common factor
The function \(g(s_i, x_t)\) acts as a multiplier on each asset's volatility. In calm times (\(x_t \approx 1\)), the multiplier is close to 1 and asset volatility is driven mainly by its own idiosyncratic shocks. During a common volatility event (\(x_t \gg 1\)), the multiplier amplifies volatility for all assets in proportion to their sensitivity \(s_i\).
COVOL-Adjusted Volatility (CAV)
While the raw COVOL factor \(x_t\) tells us whether common shocks are above or below average, it does not tell us the actual level of risk in familiar terms. CAV translates the COVOL output into an annualized volatility number by combining two ingredients: the GARCH model's baseline volatility estimate (which captures persistent, predictable risk) and the COVOL surprise factor (which captures the additional risk from common shocks):
where:
- \(h_{a,t}\) is the GARCH conditional variance for analysis \(a\). This is the model's best estimate of daily volatility based on recent history
- \(c_a\) is a scaling constant that corrects for the number of assets in the analysis (specifically, the sum of the PC1 eigenvector weights), ensuring CAV values are in meaningful units
- \(g_{a,t} = s_a(x_t - 1) + 1\) is the COVOL adjustment factor. When \(x_t = 1\) (average conditions), this equals 1 and CAV simply reflects the GARCH baseline. When \(x_t > 1\) (elevated common shocks), it amplifies volatility beyond what GARCH alone would predict
- \(252\) converts from daily variance to annualized volatility (there are approximately 252 trading days per year)
The resulting CAV values are expressed as annualized percentages, the same units used for standard volatility measures like the VIX. In normal periods, CAV values typically range from 10% to 25%. During major crises such as the 2008 financial crisis or the 2020 pandemic, they can reach 100% to 300%, reflecting the extreme common volatility experienced across markets.
Aggregate COVOL-Adjusted Volatility (ACAV)
Each of the four COVOL analyses produces its own CAV value. To combine them into a single headline risk number, we construct a hypothetical portfolio that gives equal risk contribution to each analysis and then calculate that portfolio's volatility. This approach ensures that no single analysis dominates the aggregate simply because it happens to be more volatile. Instead, each analysis contributes equally to the overall risk.
The construction proceeds in three steps. First, we assign risk-parity weights. Analyses with higher CAV get lower weight, so that each contributes roughly equally to portfolio risk:
Next, we estimate how correlated the four analyses are with each other. We do this by computing rolling correlations of their standardized innovations (the part of each analysis's returns that is unexplained by its own GARCH-COVOL model). These correlations, combined with the individual CAV values, form a covariance matrix:
where \(\rho_{ij,t}\) is the rolling correlation between the standardized innovations of analyses \(i\) and \(j\).
Finally, the Aggregate CAV is the portfolio volatility, the square root of the weighted covariance:
When all four analyses are highly correlated (as during a global crisis), ACAV will be close to the average of the individual CAV values. When correlations are low, ACAV benefits from diversification and will be lower than the average, reflecting the fact that idiosyncratic risks partly cancel each other out.
Step 1: Four Independent Stress Readings
COVOL measures are computed for four analysis groups: Country Risk (COUNTRY), Oil & Gas (OILGAS), Asset Classes (AC), and Commodities (COMM).
Each COVOL analysis estimates a common volatility factor \((x_t)\) and per-asset factor loadings \((s_i)\) for the assets in that group. Higher \(x_t\) values indicate that common volatility shocks within the group are larger than average.
Step 2: Finding the Common Thread (PCA)
Why PCA? When all four analyses spike together, something is affecting the entire market. PCA extracts this common factor (PC1) from the noise.
Principal Component Analysis finds what the four COVOL readings share. PC1 captures market-wide stress hitting all analyses simultaneously.
Loadings show how connected each analysis is to this common factor and are stored as time series, revealing how the structure of market stress evolves over time.
Step 3: PC1 Loadings Over Time
Why track loadings? The relationship between analyses and the common factor is not static. Loadings change as market regimes shift, revealing which analyses are most connected to broad stress at any given time.
Each analysis's PC1 loading measures how strongly it participates in the common factor. A rising loading means that analysis is becoming more correlated with market-wide stress. A falling loading means it is decoupling from the common factor.
Aggregate CAV (ACAV)
The Aggregate COVOL-Adjusted Volatility provides a single portfolio-level risk measure:
ACAV combines the individual CAV values from all four analyses using risk-parity weights and a covariance matrix built from rolling correlations of standardized innovations. This accounts for both individual volatilities and their correlations, providing a comprehensive view of total market risk.
When ACAV is elevated but PC1 is moderate, risk is high but not fully systematic — individual analyses are stressed independently. When both ACAV and PC1 are high, risk is elevated and highly correlated across analyses, meaning diversification benefits are severely reduced.
Interactive Features
The Stress Timeline chart offers several interactive features for detailed analysis of market stress patterns.
View Mode Selection
Click the View Options button next to the Stress Timeline header to access different visualization modes:
- Composite Overview (Default) Shows the ACAV chart on top and Systematic stress (PC1) below, with synchronized hover interactions. This is the primary view for assessing overall market risk and how much of it is driven by the common factor.
- PC1 Loadings Displays the time-varying PC1 loading for each of the four analyses. Useful for understanding how the structure of market stress evolves and which analyses are most connected to the common factor at any point in time.
- COVOL-Adjusted Volatility Shows the CAV values for each analysis as annualized volatility percentages. Useful for comparing risk levels across analyses and understanding which market segments carry the most volatility risk.
Analysis Visibility
In PC1 Loadings or COVOL-Adjusted Volatility views, toggle individual analyses on or off using the checkboxes. Each analysis has a distinct color: purple for Country Risk, amber for Oil & Gas, cyan for Asset Class, and emerald for Commodities.
Stacked Chart Option
In the COVOL-Adjusted Volatility view, enable Stacked mode to see how individual analysis contributions combine. This visualization helps understand how much each analysis contributes to overall market risk at any point in time.
Historical Data Exploration
Hover over the chart to preview historical values. Click to pin a specific date, which updates all dashboard components to show that historical snapshot. Click another date to change the selection, or use the close button to return to current data.
Stress Decomposition Panel
Below the timeline chart, on the left side
- Each metric is shown as a dot positioned on a horizontal density strip, indicating where the current value falls within its historical distribution.
- The density strip represents the full historical distribution of that metric. The dot's position immediately conveys whether the current reading is normal, elevated, or extreme.
- Individual analysis CAV values are also shown as dot-on-distribution strips, allowing quick comparison of risk levels across all four analyses.
Analysis Contribution Radar
Below the timeline chart, on the right side
- Four axes represent each analysis: Country (purple), Oil & Gas (amber), Asset Class (cyan), and Commodities (emerald).
- Distance from center shows each analysis's PC1 loading magnitude. The furthest axis indicates which analysis is most connected to the common factor.
- All axes use the same scale based on a 252-day rolling window for fair comparison.
- Hover over dots or legend items to highlight specific analyses.
Data Download
Download button in the widget header, next to the title
Click the Download button to export the complete COVOL Composite Index time series as a CSV file. The download includes:
Date, Systematic (PC1), Aggregate CAV, Country CAV, AC CAV, OILGAS CAV, COMM CAV, Country Loading, AC Loading, OILGAS Loading, COMM Loading.
Data Interpretation
The COVOL Composite Index helps identify market risk regimes and understand whether current conditions are driven by a shared systematic factor or elevated risk in individual analyses.
Understanding Percentile Ranks
The percentile shows how current stress compares to historical values:
- 0-25% (Low risk): Markets calm, diversification working normally.
- 25-50% (Below average): Normal market conditions.
- 50-75% (Above average): Moderately elevated, review hedging positions.
- 75-90% (High risk): Significant pressure, active risk management recommended.
- 90-100% (Extreme): Crisis-level conditions, expect correlation breakdown and reduced diversification benefits.
Understanding the Decomposition
The Stress Decomposition visualization uses dot-on-distribution strips to show where current readings fall within historical context. The ACAV strip shows overall portfolio risk, the PC1 strip shows systematic stress, and individual analysis strips show per-analysis CAV values. Together, these provide an immediate visual read on whether risk is normal, elevated, or extreme across all dimensions.
COVOL Analysis Contribution
The COVOL Analysis Contribution radar chart reveals which market segments are most connected to the common factor. The analysis with the highest PC1 loading is most correlated with systematic stress, while analyses with lower loadings are more independent.
Common Patterns
Here are typical patterns you may observe and what they indicate:
High PC1, high ACAV, all loadings elevated
Meaning: Pure systematic stress, with all markets moving together in sync and aggregate risk at extreme levels.
Response: Broad de-risking recommended; analysis-specific hedges will not help since everything is correlated.
Moderate PC1, high ACAV, one analysis CAV much higher than others
Meaning: Aggregate risk is elevated but driven primarily by one analysis rather than broad correlation.
Response: Analysis-specific hedges may be effective; investigate that analysis's fundamentals.
High PC1, one analysis with low loading
Meaning: One analysis is decoupled from the common factor, behaving independently during market-wide stress.
Response: Consider that analysis's assets for tactical allocation during the stress period.
Practical Applications
Risk Management
Monitor ACAV for aggregate portfolio risk, track PC1 for systematic stress levels, and use per-analysis CAV and loadings to understand risk concentration and correlation structure.
Market Timing
Use the ACAV and PC1 relationship for entry/exit timing, identify when diversification benefits are diminishing via rising loadings, and recognize rotation opportunities when individual analysis CAV values diverge.
Academic Research
Study stress transmission across analyses via time-varying loadings, analyze the relationship between aggregate and systematic risk, and research volatility spillover mechanisms using CAV decomposition.
Understanding Data
The COVOL Composite Index is computed daily using the four COVOL analyses maintained by V-Lab. Data is available from 2000 to present.
Analysis Coverage
The index covers four major COVOL analyses: Country Risk (COUNTRY) covering country-level market exposures; Oil & Gas (OILGAS) covering energy sector equities; Asset Classes (AC) covering diversified asset class exposures; and Commodities (COMM) including commodity futures and related equities.
Troubleshooting
Common Questions
Why does the systematic stress (PC1) sometimes spike when equity markets appear calm?
The systematic component monitors stress across multiple analyses including commodities, oil & gas, and country risk. Stress events in these analyses may precede or occur independently of equity market volatility, providing early warning signals.
What is the difference between ACAV and PC1?
ACAV (Aggregate CAV) measures the total portfolio-level risk by combining individual analysis CAV values using a risk-parity variance formula. PC1 measures how much of that risk is shared across analyses. High ACAV with high PC1 means high correlated risk. High ACAV with moderate PC1 means high but less correlated risk.
What does it mean when one analysis has a much higher loading than others?
A high PC1 loading means that analysis is strongly connected to the common market factor. This analysis will move most closely with systematic stress. Conversely, an analysis with a low loading is more independent and may offer diversification benefits during stress periods.
User Questions from V-Lab Community
Based on technical questions from V-Lab users:
How are the four analyses weighted in the systematic (PC1) calculation?
The analyses are not explicitly weighted. Instead, PCA automatically determines each analysis's contribution through its loading on PC1. Analyses that move more closely with the common factor receive higher effective weights. These loadings are stored as time series, so you can track how the weighting evolves over time.
How should I interpret historical stress levels compared to current readings?
The dot-on-distribution strips in the decomposition panel provide immediate visual context. Each strip shows the full historical density, and the dot marks the current value. Focus on where the dot falls relative to the distribution rather than absolute values, as market structure evolves over time.
Tips & Best Practices
Monitor the ACAV value for overall risk level and the PC1 value for how much of that risk is systematic. Use the decomposition strips and loadings radar to understand the structure of current market conditions.
Analysis Workflow
1. Check the ACAV and PC1 values in the header. ACAV provides the headline risk level; PC1 shows how much is systematic. 2. Look at the default dual-chart view. The top chart (ACAV) shows aggregate risk over time; the bottom chart (PC1) shows systematic stress. Hover to compare them at any date. 3. Check the decomposition panel's dot-on-distribution strips. Dots near the right tail indicate unusually elevated readings. 4. Check the radar chart for PC1 loadings. The furthest axis shows which analysis is most connected to the common factor. 5. Switch to the COVOL-Adjusted Volatility view to compare individual analysis risk levels. 6. Switch to PC1 Loadings view to see how each analysis's connection to the common factor has evolved. 7. Click on the timeline to compare current readings with historical crisis periods. 8. Drill down into individual COVOL analysis pages to examine the underlying assets, top events, and detailed volatility dynamics for any analysis that stands out.
Was this page helpful?