Can anybody give me some insight as to how firms control risk across multiple asset classes? Do they typically hire a quant/risk specialist to build out software and monitor? Or are there canned products that can be bought and or customized to meet a firms criteria? Thanks in advance.
At the most fundamental level it involves finding a way to compare both risk and reward across all classes, for example the sharpe ratio (or sortino or others), there are also other alternative measurements. This gives a risk adjsuted return, and then the allocations are adjusted so as to control risk within allowable parameters. There are flaws in this approach, largely to do with the measurement of risk and having to rely on past results
I was hoping to hear a few answers from the trenches myself. I think that there are a variety of answers in use. Here is what I know: We wrote a custom OMS (order management system) / Trading Accounting systems for a CTA once. We did it his way. A small mostly-futures Hedge Fund that I knew pretty well depended upon Risk Metrics to get them up and running on a SaaS (Software as a Service.) I would think that it would matter depending upon how big you are. Are you trying to wow your uncle by saying that you use VAR? Are you obligated under Basel/Basel II? On the job boards, I see that people are building their own stochastic integration (Monte Carlo) VAR engines. Given the model construction dependency of factor VAR methods, and given the rather peculiar times in which we trade, I suspect that people are going Monte Carlo to avoid this. For canned software, take a look at the names at http://software.nasdaq.com/risk-management-software There are many more entrants than that. Since portfolio allocation doesn't occur in a vacuum, look at the other pieces that you will be needing. For example, Charles River Development or Eze Castle Software. If you really just want risk numbers, I would talk with RiskMetrics about getting a starter-rate. They will get you going.
hello, sorry that I'm also posting here but I already posted in another section but there wasn t to much interest, probably because that VaR topic is in the "options" section. I would like to ask for help about variance-covariance VaR calculations. I found a interesting video about how to calculate VaR in excel for two assets but I don't really know in details how to calculate it for more assets( 3 and more assets ). youtube.com/watch?v=YR2MijzLRPo Can anyone help me with a explanation about how to calculate VaR(for linear instruments) for more assets? Some formula or other indications. Thank you in advance.
Doing a meaningful job at Value-at-Risk depends critically on having meaningful correlation and volatility estimates. These are not easy to do well. I suggest getting a book by Carol Alexander. I have found, "Market Models: A Guide to Financial Data Analysis" to be invaluable, but she has written several since then.
Thanks for the replay. For the moment I just want to start with that simple example but the problem is that my math knowledge are very low and I don't know how to calculate that version of VaR based on that example.
Anyone?... I found this example jpmorgan.com/tss/General/Risk_Management/1159360877242 At example 4, they gave an example on how to calculate for 2 assets. Could that formula be modified to include 3 assets for instance? I was thinking to extend it to something like this for 3 assets: w1 s1 + w2 s2 + w3 s3 + 2 w1 w2 w3 s1 s2 s3 P It is correct?