Is VAR a good measure if the portflio includes short positions?
Kindly help.
Thank you
yes, VAR= portfolio (er -(1tail *std dev)
I’m not so sure about that. VAR assumes normal distribution and long-short portfolios, I think, do not exibit normal distribution of outcomes. Can someone else confirm that?
var assumes normality especially var / cov method; i dont think VAR would be appropriate in the case of short positions.
Not good, because of normality assumption. Monte Carlo can be used for portfolios with embedded options.
If the portfolio includes bothlong and short (like short extension strategy) or market neutral strategy (Beta =0), then can we apply VAr as a risk measure.
thank you
Sorry, I misread. I also think that if portfolio includes short positions parametric VAR shouldn’t be used.
what’s the source?
Kobi, source - I did not understand
I was referring to the two strategies included in the equity portfolio management.
So can VAR be included as a risk measure for these two equity portfolio management strategy.
Thank you
To conclude, can VAR be used when short positions are included in a portfolio?
I would suggest the VAR risk measure can certainly be used on a long/short portfolio…but with the provision that both your long and short positions be expressed in the “absolute value” of their financial stake/outcome as a netting of both positions would distort VAR by failing to express the true absolute invested position. Thats assuming the “short” is not executed through options as that would introduce non-normality which renders VAR less representative.
Interesting question though especially if the short position contracts can reflect a shorter timeline than the long positions. Not sure what the “CFA” answer would be!
Actually the short positions (sell the security up front and commit to sell it back at specified time in the future) introduces leverage which can make the returns more volatile but I dont think it breaks the normality assumption of VAR
thats buy it back at specified time in future!
yikes, I couldn’t find this in the notes and a quick search of the web gives conflicting answers.
Is it in the cirriculum?
No, I did not come across in the schweser note. This is coming out of my thought.
Certainly the energy and motivation to think, largely is attributed to this forum.
I studied very studiously from Dec 2013. I am not very strong in risk management that is to say, if I a tough question is put to me, I will scrible and prepare for the next time.
One of my recent repetences in life is that I have not know about this forum for a very long time.
I came across this site in google search for roll return, that is about a month ago
I hate to admit my lacksity in the preparation
I would think VAR can be used for long/short. A short position has symmetrical returns, just like a long. And although it does introduce the dynamic of leverage, being levered amplifies returns on both the upside and downside (I believe equally). So, normality probably not violated. Options on the other hand…different story.
Though it does beg the question - what if the portfolio includes positions (long or short) in derivatives other than options, like forwards/futures/swaps? I’d imagine caps/floors follow the same pattern as options - not sure though…
if leverage makes the lower tail fatter, then this would probably violate normality.
Long and short positions are okay with VAR however if you add put; then normality assumption doesn’t hold anymore.
Even long short normally won’t be Normally distrubed. Say you have a good manager with Average Alpha of one. Max Alpha of 2 and Max lost of -2. If they mostly get Alpha (average/mean = 1). The skewed to the left, which destroys the normality assumption.
Short positions and puts are two different things. Puts have asymetric payoffs since there is no downside risk once purchased other than loss of the principal. Short positions are just the opposite of long positions so they should exhibit normality. Leverage would imply more volatile returns but whether it violates the kurtosis issue would depend on the calculation of kurtosis and the excess kurtosis, 3 is standard I think. Then the area under the curve calculated using standard deviation wouldnt be accurate.