On this page
Inspect the supplied 90% example
The CSV uses the original loss distribution and records cumulative probabilities, available mass, selected tail mass and each selected probability-times-loss term. The example is static teaching data rather than an automatically recalculating spreadsheet.
Reconcile the selected mass
At 90% confidence, selected probabilities are 0.05 at loss 50 and 0.05 at loss 10. Their sum is 0.10. Their weighted losses sum to 3, so ES is 30. Other outcomes receive zero inclusion mass.
Review the allocation constraints
| Check | Required relationship |
|---|---|
| Available mass | Each original probability is nonnegative |
| Included mass | Between zero and available probability |
| Selected sum | Exactly the chosen worst-tail probability |
| Priority | Worse losses allocated before smaller losses |
| Average | Selected weighted loss / selected probability |
Retain a fresh model record
Changing an outcome or confidence requires a new allocation, not just a new heading. Keep the original model probabilities and units next to the bridge so a threshold subset cannot be mistaken for the selected tail.