Vince,

I was thinking that max (negative)  potential exposure for different counterparties could occur for different scenarios and so combining max potential exposure across counterparties may overstate cash needs.

If overall scenario for overall max (negative) potential exposure is considered, then we will pick up some high exposures and some low exposures, which may be fine.

Alternately, why not tweak the potential exposure program to do deterministic shifts of one sigma and two sigma and assuming one factor model and perfect correlation across commodities.
Then the negative mtm values can be compared to collateral already outstanding.
This method does not guarantee that we will get large exposures, but is easy to explain.

Another thing to check with Winston is that the program does not automatically zero out negative exposures. If it does, then we may have to flip the sign for all positions, get the exposures, then flip back the sign. Then compare to collateral outstanding.

Vasant
--------------------------
Vasant Shanbhogue