[R-SIG-Finance] [R-sig-finance] Commodity swap?

[Brian G. Peterson]

Bogaso wrote:
> unlike most interest rate swaps explained in common risk management book, I
> found that in most of the exchanges like IPE etc, cash flow is generally
> happen daily (where for interest rate case it is mostly monthly or quarterly
> etc) and all of the case underlying is some futures contracts. 
> 
> In ordinary interest rate case future expectation is generally replaced by
> the futures quote. Therefore my question is, if swap is based on futures
> itself then how can I get unbiased expected value as proxy?
> 
> Or should I treat this kind of swap contract just like a Basis wherein Basis
> = (tomorrow's future quote - fixed) and try to understand the tomorrow's
> possible distribution that future quote and hence the VaR just the same way
> as Delta-normal approach?
> 
> What I mean is that :
> 
> VaR in this approach :
> 
> 5th worst of (Basis[t+1] - Basis[t]) = (Futures[t+1] - Futures[t]), as other
> leg is fixed and therefore no risk is there.
> 
> Is it the correct? Then how should I incorporate term structure which is
> generally the case for interest rate swap?
> 
> If somebody can give some view, it would be great. 

Christofer,

Every swap contract is different, and you didn't actually tell us what contract 
you're looking at.  So replies will of necessity be somewhat more general than 
your general questions...

Considering a futures contract to be an 'unbiased expected value as proxy' is a 
bit of a stretch in any event.  Futures, like any other financial instrument, 
exhibit all sorts of biases and market forces. All that aside, you're on the 
right track.

The risk of *any* swap is the risk of the underlying basket portfolio.

Given that you can model a swap as a basket portfolio, you are correct that you 
would model the risk of the underlying futures contract.  However, I don't 
understand why you are thinking about one-day risk as opposed to some longer 
time horizon.  In some frameworks (e.g. Basel II, or a day trading firm) daily 
risk is required, but in many investment frameworks, you're most concerned with 
longer time horizons, even if the contract is valued daily.  It is of course 
straightforward to aggregate prices or returns to a longer horizon.  (using 
to.period in xts for price series or PerformanceAnalytics' portfolio return 
functions for return series in R)

You would model the term structure using any of the many tools available in R 
to do so (termstruct comes immediately to mind, though there are many other 
models represented in other packages). This would enable you to add interest 
rate basis risk to your market risk estimates.

Any complex synthetic instrument may be modeled as though it were a portfolio 
of the underlying or representative assets.  What you choose as your risk 
measure (Expected Shortfall, Delta-VaR, Conditional expectation of Drawdown, 
etc.) is, as always, a business decision based on your investment style and 
time horizon.

Regards,

    - Brian

-- 
Brian G. Peterson
http://braverock.com/brian/
Ph: 773-459-4973
IM: bgpbraverock