On this page
Identify the risk question
A loss quantile asks where the model’s cumulative probability reaches a confidence level. Expected shortfall asks for the average loss of exactly the worst complementary mass. Use the same loss convention, horizon and model when comparing them.
Read the original outputs
| Summary | Loss units | Interpretation |
|---|---|---|
| Lower VaR | 10 | First cumulative crossing |
| Expected shortfall | 30 | Worst 10% weighted average |
| Maximum listed loss | 50 | End of finite model support |
| Whole-model mean | 2 | Average over all probability |
Keep tail shape available
Models can share a VaR threshold while differing in losses beyond it. ES responds to the selected tail amounts, but remains dependent on supplied probabilities and outcomes. Neither statistic captures all liquidity, operational or model uncertainty.
Report parameters with the number
Retain confidence, horizon, unit, sign convention and estimation basis. A lone number labelled risk cannot establish which of these questions it answers. Adding a course link or a chart does not repair a missing model record.