Delta Normal VaR
One of the simplest ways to calculate value at risk (VaR) is to make what are known as delta-normal assumptions. For any underlying asset, we assume that the log returns are normally distributed, and we approximate the returns of any option based on its delta-adjusted exposure. For portfolios, the delta-normal model assumes that the relationships between securities can be fully described by their correlation.
The delta-normal assumptions make it very easy to calculate VaR statistics even with limited computing power. This made delta-normal models a popular choice when VaR models were first introduced. Predictably, the results of such a simple model were often disappointing. Delta-normal models are rarely used in practice today, but are still an excellent starting point when learning about VaR models. By understanding the pros and cons of the delta-normal model we will be better able to understand the pros and cons of more complex models. Unfortunately many people outside of risk management believe that delta-normal models are still widely used in practice, or believe that the shortcomings of these simple models are somehow inherent to all risk models.
For more information on VaR models, see our white paper, An Introduction to Value at Risk.