I am running a VaR calculation and have seen two ways of doing it in several places online.
One simply assumes normal distribution of returns and selects n number of returns from the normal distribution.
The other again assumes normality and uses Geometric Brownian Motion as the process to get the returns.
I am assuming here that both methods are a random walk but simply with a different underlying process?
What’s the downfall with just simply selecting from the normal distribution as opposed to GBM, also isn’t selecting from the normal distribution in this way going to just yield the same result as the parametric VaR with enough simulations?