Conditional Value-at-Risk (CVaR)

The precursor to CVaR, VaR estimates the minimum loss expected at a specific confidence level (e.g., 95%) over a defined horizon. Unlike CVaR, VaR is not a coherent risk measure—it fails the sub-additivity property, meaning diversification can appear to increase risk. VaR also remains silent on the magnitude of losses beyond its threshold, a critical blind spot that CVaR directly addresses by averaging the tail.

Synonymous with CVaR in continuous distributions, Expected Shortfall is the formal term under Basel III regulations. ES calculates the probability-weighted average loss in the tail beyond the VaR cutoff. Key properties:

• Coherent: Satisfies monotonicity, sub-additivity, homogeneity, and translational invariance
• Spectral: Weights all tail losses equally, unlike VaR which ignores tail shape
• Regulatory standard: Replaced VaR for market risk capital calculation under Basel III's Fundamental Review of the Trading Book (FRTB)

The practical application of CVaR insights—constructing positions that pay off during extreme events. A portfolio with high CVaR signals the need for convex hedges. Common implementations:

• Long OTM puts: Direct protection against equity crashes
• Variance swaps: Pure volatility exposure without delta path dependency
• Tail risk funds: Dedicated strategies harvesting crisis alpha CVaR optimization often reveals that a small allocation to deeply convex instruments can dramatically reduce expected tail losses without sacrificing mean returns.

The statistical foundation for estimating CVaR when historical data is sparse. EVT models the asymptotic behavior of tail distributions using the Generalized Pareto Distribution (GPD) for peaks-over-threshold approaches. Key techniques:

• Block maxima: Fits GEV distribution to maximum losses per period
• Peaks-over-threshold: Models all exceedances above a high threshold
• Hill estimator: Estimates tail index for power-law distributions EVT enables CVaR calculation at confidence levels (99.9%) where empirical data provides zero observations, critical for stress testing and regulatory capital.

A complementary risk metric to CVaR that measures the largest peak-to-trough decline in cumulative returns. While CVaR quantifies expected loss severity in a single period, MDD captures the path-dependent pain of holding through a prolonged decline. Key differences:

• CVaR: Forward-looking, parametric, single-period expected tail loss
• MDD: Backward-looking, empirical, multi-period worst-case realized loss Institutional allocators often constrain both: CVaR for regulatory capital and MDD for redemption risk management, as large drawdowns trigger investor withdrawals.

The non-parametric complement to CVaR models. While CVaR relies on distributional assumptions, stress testing simulates portfolio losses under specific historical or hypothetical scenarios:

• Historical: 2008 GFC, 2020 COVID crash, 1987 Black Monday
• Hypothetical: Sovereign default cascades, cyber-attack market closures
• Reverse stress testing: Identifies scenarios that would breach CVaR limits Regulators require both CVaR-based models and scenario analysis under CCAR and DFAST frameworks, recognizing that tail dependence structures break during crises.