[R-SIG-Finance] [R-sig-finance] A VaR question

[Cedrick W. Johnson]

<warning: coffee-has-not-kicked-in-yet>
A chime in regarding the calculation of VaR on spreads. I follow the 
approach Brian mentioned below, in addition to calculating the VaR for 
the actual  spread itself which yields yet another metric called 
'Diversification Benefit' (subtracting the spread VaR minus the combined 
VaR of the two legs). I'm trying to recall amongst the stacks of papers 
how to determine VaR on spreads that could be <= 0 (I admit, I've taken 
a rather haphazard approach and eliminated the timeseries minus the 0/- 
periods in question).

-cj



Brian G. Peterson wrote:
> I calculate the risk of a particular spread using the underlying 
> assets as you suggested in 1>
>
> However, your risk if you are short the spread is (potentially) 
> different than your risk if you are long the spread, so keep that in 
> mind as well.
>
> Another approach is to calculate the portfolio risk of the trading 
> strategy, regardless of the underlying assets, but you need trade and 
> P & L history for that approach.
>
> Regards,
>
>     - Brian
>
> "Bogaso" <bogaso.christofer at gmail.com> wrote:
>
>> Hi all,
>>
>> My question is not directly R related but rather a finance related 
>> question.
>> Therefore I was wondering wheher I find a reliable answer here.
>>
>> Here I wanted to calculate VaR for basis (spot-future). There could 
>> be two
>> approaches : 1: Assuming basis as a portfolio of two assets and then
>> calculate the risk of the spread, 2 : Create a historical price 
>> series of
>> basis then calculate VaR like single asset portfolio.
>>
>> Which one would be correct approach? In my opinion 1st is correct 
>> because,
>> as basis can get any value like +ve & -ve, cashflow is not well 
>> defined in
>> the sense that, if I sell basis (as an asset) and that time basis is
>> negative, then I actually paying money for selling my asset !!! and 
>> secondly
>> I cannot calculate percentage/logarithmic return for basis as basis 
>> can take
>> zero-value as well.
>>
>> Can anyone validate that? What is the standard approach for 
>> calculating risk
>> of a spread series? Should not we consider the fundamental risk factors
>> (like in basis-case they are spot & future)?
>>
>> Best
>> -- 
>
> -- 
> Brian G. Peterson
> http://braverock.com/brian/
> Ph: 773-459-4973
> IM: bgpbraverock
>
> _______________________________________________
> R-SIG-Finance at stat.math.ethz.ch mailing list
> https://stat.ethz.ch/mailman/listinfo/r-sig-finance
> -- Subscriber-posting only.
> -- If you want to post, subscribe first.