Source: Jorion, P (2007), Value at Risk: The New Benchmark for Managing Financial Risk (3rd ed), New York: McGraw-Hill.
See also:
VaR Methodologies Compared
The Pico/W In C: Servos
36 minutes ago
Thursday, March 18, 2010
VaR Methodologies Compared II
In addition to the Value at Risk (VaR) comparison chart I posted at VaR Methodologies Compared, I found a second equally insightful comparative analysis in table form, this one by Prof Philippe Jorion (2007, p. 270):
As stated in my previous post, I believe that Monte Carlo (i.e., stochastic) simulation provides the most robust approach to estimating VaR, though this approach requires more computing (and brain) power.
Source: Jorion, P (2007), Value at Risk: The New Benchmark for Managing Financial Risk (3rd ed), New York: McGraw-Hill.
See also:
VaR Methodologies Compared
Source: Jorion, P (2007), Value at Risk: The New Benchmark for Managing Financial Risk (3rd ed), New York: McGraw-Hill.
See also:
VaR Methodologies Compared
1 comment:
I fully agree. In my experience conditional VaR (Expected Shortfall) from a Monte Carlo system is the best approach to estimate the risk on non-linear products. There are inherent shortcoming in historical simulations as applied to any class of products, while the delta-normal may suffice only for a limited subset (where normality and zero 2+ moment assumptions hold).
Any approach is as good as the model and, in reality, I don't believe that a single number (VaR or ES) can convey the risk of a complex portfolio. An array of sensitivities is a must and, to use a cliché, a picture is worth a thousand words, but only to the informed observer. Interpreting a portfolio P/L plot of MC realizations still requires a solid understanding of the products, market mechanics, current economic conditions and, last but not least, the model assumptions/limitations.
To use a favorite quote of a fellow coworker: "Everything should be made as simple as possible, but not simpler." Einstein
Post a Comment