Value at Risk is useful because it compresses a loss distribution into one threshold. Expected Shortfall goes further: it asks what the average loss looks like once that threshold has already been breached.

01 / Why VaR is incomplete

VaR answers a quantile question. At 99%, it estimates a loss level that should only be exceeded 1% of the time under the model. The problem is that it says very little about the severity of the losses beyond that point.

02 / Expected Shortfall as stress language

Expected Shortfall focuses on the tail beyond VaR. This makes it more useful when the portfolio contains nonlinear exposures, crowded trades, liquidity gaps or option-like payoffs where extreme outcomes matter more than the threshold itself.

03 / Practical modelling note

The statistic is only as good as the scenario engine behind it: window length, volatility regime, correlation assumptions, liquidity horizon and backtesting discipline matter. A clean dashboard should show the number, the assumptions and the failure modes together.

The objective is not to make risk look mathematically impressive. It is to make uncertainty visible enough to be acted on.

Reference note: Basel Committee market risk framework.