Sungard principal weighs up pros and cons of the popular value-at-risk (VaR) metric tool
The Value at Risk (VaR) simulated paths, which is a popular metric used in risk management, while having low probability, have the potential to cause serious harm to trading operations says one of Sungard’s principals.
“While VaR gained popularity in the 1980s after the market crash of 1987, today it is being scrutinized after the market crash of 2008 due to extreme tail events that were not properly identified or mitigated,” says John Doran, Principal at Sungard Global Services Energy Practice. “Visibility into the behavior of the tail or the distribution of the simulated paths is generally not communicated to the stakeholders that use the metric.”
However, Doran says that new visualization techniques are emerging that make it possible to increase transparency around risk, enhance process efficiencies, and boost revenue opportunity.
VaR, which calculates the risk associated with open positions based on historic volatility, holding period and given confidence levels has been criticised by some of the biggest names in the industry.
VaR models are dangerous, said Anthony Ward, Co-Founder of Armajaro, where he specialises in soft commodities trading, when speaking at the GAIM Conference in 2011. VaR is no good for anyone and the only reason we use it is because everyone keeps asking for it. We invest in options to reduce volatility, which is another tool we can use to manage risk.
At the same conference Nassim Taleb, Principal, Universa Investments, distinguished Professor of Risk Engineering and co-Director of the Research Center for Risk Engineeering at the Polytechnic Institute of New York University and author of the Black Swan also expressed his dislike for VaR.I have been trying most of my life to explain that traditional risk models that people use do not work, said Taleb. And obviously, we now have evidence that those who dont have those models, have better sense of heuristics and intuition than those who do. VAR doesnt work, never works and never will work.
On the conferences opening day, he revealed his Heuristics formula. The Heuristics formula is computing the model or risk model at parameter p. The formula itself is p+x%, p-x% (where x% is the average deviation.)
The model or measure, includes the stress test (p is the stress test, say -15%, -20%, -25%)
Doran says that there are two main challenges to VaR.
“One of the primary problems with VaR is the fact that it is just a single number,” says Doran. “Nothing about the behavior on the extreme tail or the distribution of the simulated paths is generally communicated to the stakeholders that use the metric.The other major challenge with VaR is the fact that it is non-additive. VaR can be broken down into component levels, but one cannot add and subtract these values. However, detailed simulation trials from Monte Carlo can be added and subtracted from one another. This is not commonly done today, since the volume of data would exceed the threshold with which most organizations are comfortable working. Methods and capabilities associated with large data sets (big data) are beginning to gain momentum. In fact, some of the techniques available today can be applied to the simulation results from Monte Carlo. Using advanced visualization techniques, one can begin to observe and analyze VaR in new and unique ways.”
(LB)