Definition
Value at risk at confidence level alpha is a loss quantile: a threshold that is exceeded with probability no more than 1 minus alpha under the chosen loss distribution and horizon. Its sign and quantile convention must be stated because some systems report a positive loss and others a return quantile. VaR says little about the severity of losses beyond the threshold.
How it is calculated
Value at risk at alpha equals the alpha-quantile of the loss variable L over the stated horizon.
| Symbol | Meaning | Unit |
|---|---|---|
| VaR | value at risk | loss units |
| alpha | confidence level | probability |
| L | loss variable | currency or return loss |
| q | quantile function | loss units |
Interpretation
VaR draws a line in the loss distribution: most outcomes are on the less severe side under the model. It does not tell you how far the worst outcomes extend. Two portfolios with the same VaR can have very different tails.
Assumptions and conventions
-
The loss convention, horizon, confidence level and quantile definition are explicit.
-
The confidence level satisfies 0<alpha<1.
-
The loss distribution is estimated or modeled from an appropriate information set.
Loss units over the stated horizon; alpha is dimensionless.
Limitations and common misreadings
-
VaR is not the maximum possible loss.
-
VaR alone is not expected shortfall.
References
- Alexander J. McNeil. (2005). Quantitative risk management: concepts, techniques, and tools. Chapter 2, Basic Concepts in Risk Management. isbn:9780691122557. Source

