Model risk
The risk that the model behind a decision is itself wrong: assumptions do not hold, parameters were fitted, a formula is used outside the range where it works.
How to read the number
The more elaborate the construction, the more convincing its output looks and the harder it becomes to notice that the inputs were chosen arbitrarily.
When the metric lies
A model calibrated on a quiet stretch breaks precisely where it is needed most: in a crisis the relationships between assets differ from the sample it was tuned on.
Also known as: modelling error risk