Skip to content

Exact cumulative boundary in a loss quantile

In the original model, the lower 95% loss quantile is 10 because cumulative probability at 10 is exactly 95%.

On this page

Retain the cumulative probabilities

Original same-horizon loss model
Loss unitsProbabilityCumulative probability
−1030%30%
040%70%
1025%95%
505%100%

Locate the first value satisfying the rule

At loss 0, cumulative probability is 70%, below 95%. At loss 10 it is 95%, which satisfies F(l)≥0.95. Therefore the lower quantile is 10. Selecting the next row would instead apply a different strict-crossing rule at this boundary.

Calculate the selected 5% average

At 95% confidence the worst-tail mass is 5%. All of it is supplied by loss 50. Expected shortfall is (0.05×50)/0.05=50. Probability strictly above VaR 10 is 5%; probability at or above it is 30%.

Keep quantile and severity separate

VaR 10 and ES 50 are consistent. The threshold describes a cumulative crossing; ES describes the selected worse tail’s average. Neither number is a promised limit on future loss.

Further references