Expected shortfall is a tail risk measure that asks a different question from value at risk.

Expected shortfall is a tail risk measure that asks a different question from value at risk. Value at risk identifies a loss threshold at a chosen confidence level. Expected shortfall then looks only at the cases where that threshold is breached and takes the average loss within that tail. In plain terms, value at risk says where the tail begins, while expected shortfall says how deep the losses are once the portfolio is already in that tail.

In the Sonar fundamentals material, value at risk is described as a downside risk metric that estimates the maximum expected loss over a specified period at a given confidence level, and expected shortfall is described as a downside risk measure that estimates the average loss expected in the worst cases beyond the value at risk threshold. That definition is the core reason expected shortfall is often viewed as more informative for tail events. Two portfolios can share the same value at risk and still have very different losses once the threshold is crossed. Expected shortfall distinguishes those cases because it aggregates the severity of losses beyond the cutoff rather than stopping at the cutoff itself.

Mechanically, the construction is conditional. First choose a confidence level and compute the corresponding value at risk threshold for the loss distribution. Then isolate the observations or distribution mass with losses greater than that threshold. Expected shortfall is the conditional mean of loss in that subset. Written conceptually, expected shortfall equals the expected value of loss given that loss exceeds value at risk. This conditional averaging is what makes it a tail measure rather than a simple quantile measure.

That distinction matters because value at risk leaves open the question of what happens after the threshold is breached. A quantile can tell you that losses worse than a given level occur with some small probability, but it does not summarize how severe those worse outcomes are. Expected shortfall fills that gap by pooling the tail observations and converting them into one average tail loss figure. For risk managers, that makes it useful when the concern is not only the probability of entering the tail but also the expected scale of losses once there.

The same Sonar fundamentals source also lists expected shortfall alongside other downside-sensitive metrics such as Calmar ratio, Sortino ratio, sterling ratio, tail ratio, ulcer index, and value at risk. That placement is consistent with its role in focusing attention on adverse outcomes rather than symmetric variability. Standard deviation treats gains and losses as deviations from the mean. Expected shortfall instead concentrates specifically on the worst-loss region of the distribution.

Covered in depth in the Strategy research fundamentals pillar hub.