Conditional Value at Risk (CVaR)
Conditional Value at Risk (CVaR), also called Expected Shortfall, measures the average loss an investor expects to suffer in the worst-case scenarios beyond the Value at Risk threshold, capturing the severity of tail losses rather than just their probability.
Value at Risk (VaR) answers one question: what is the maximum loss I would expect to experience with a given level of confidence over a given period? A 95% one-day VaR of $1 million means there is a 5% chance of losing more than $1 million in a single day. What VaR does not answer is: if that threshold is breached, how bad does it get? That is the question CVaR is designed to answer.
CVaR takes the average of all losses that exceed the VaR threshold. Using the same example, if the worst 5% of daily outcomes have an average loss of $2.5 million, the 95% CVaR is $2.5 million. This distinction is critical during crises. The losses that exceed VaR are not uniformly distributed just beyond the threshold — they include outliers of devastating magnitude, and their average can be dramatically higher than the VaR level.
Regulators increasingly prefer CVaR over VaR precisely because it penalizes strategies that produce rare but catastrophic outcomes. Under the Basel III framework for bank capital requirements, the shift from VaR to Expected Shortfall was finalized in the Fundamental Review of the Trading Book (FRTB), acknowledging that VaR was too easy for trading desks to game by concentrating risk in the tail beyond the threshold without increasing reported risk.
For portfolio managers, CVaR is particularly useful for evaluating strategies with negative skewness and fat tails — exactly the strategies where VaR is most misleading. Comparing CVaR across strategies at the same VaR level reveals which strategies have more dangerous tail behavior. CVaR is also sub-additive, meaning CVaR of a combined portfolio is always less than or equal to the sum of individual CVaRs — a mathematically important property that VaR does not consistently satisfy.
Portfolio optimization using CVaR as the risk objective rather than variance or standard deviation produces different portfolio constructions. CVaR optimization tends to penalize fat-tailed strategies more heavily and can lead to meaningfully different asset allocations, particularly in portfolios that include derivatives, emerging market assets, or credit instruments prone to gap-down behavior.