On this page
State the original loss law
Assume mean loss −30 and dispersion 10, in one common unit and horizon. This normal law is centered on a gain under the positive-loss convention, but it has unbounded support in both directions.
Read the 95% quantities
Model VaR is -13.551464 and ES -9.372872 loss units. Both are negative because the selected 95% threshold and worst-5% average remain on the gain side.
Check the positive-loss event
A positive loss requires Z>3 because L=−30+10Z. Standard normal upper-tail probability at 3 is approximately 0.001349898, or 0.134990%. That is small but positive under the stated law.
Keep this distinct from an all-gain finite model
A finite record containing only negative loss amounts has no positive loss within its supplied support. The positive-dispersion normal model here does permit positive losses. A negative summary alone cannot establish which support assumption applies.