Vlady, here are my comments:

1. NG Correlations : 
- leave existing methodology while using more recent data (1 month instead of 
3 months). This should go anong with running the calibration code more 
frequently 
(at least once a week);
- put the version with weighting for correlation into production. 
2. Implement term structure of correlations for selected groups of 
commodities through joint factor loadings estimation for each of these groups 
I experimented with a group of Brent and 61NY, Christian is working with WTI, 
Brent, C3GC. The experiments should identify suitable groups. 
IT needs to provide a column in some rms table to identify the groups and 
slightly modify vatrfacs code. This will address power correlations as well.
3. FFvols: Smoothing methodology for implied vol curves corresponding to each 
of 12 months
 - develop solutions for improving the existing methodology (Research)
 - implement the proposed solutions (Research, IT)
 - test the results (Research, RAC)
4. Jump Diffusion - NG: finding the evidence of jumps, not obvious.
5 & 6. - I agree.
7 IR & FX Var in RiskTrack - should be probably #1 in this list. 
 - the VAR code is ready, 
 - some results for IR VAR are validated already. (IT, Research, Risk 
Control) 
 - some minor modifications to calibration code vatrfacs are required 
(related to FX data) (IT);
 - data completeness and consistency is still a problem, specifically: 
Futures rates (Infinity, IT);
 - FX books are not set up properly (Risk Control, IT)

Tanya.


   
	
	
	From:  Vladimir Gorny                           11/09/2000 10:41 AM
	

To: Tanya Tamarchenko/HOU/ECT@ECT, Naveen Andrews/Corp/Enron@ENRON, Wenyao 
Jia/HOU/ECT@ECT
cc: Ted Murphy/HOU/ECT@ECT, Vince J Kaminski/HOU/ECT@ECT, Debbie R 
Brackett/HOU/ECT@ECT 
Subject: VaR questions

Consistent with our research lunch discussion yesterday, I would like to 
formalize and prioritize VaR related developments. Please review the projects 
and priorities and make comments/changes. If we are in agreement, I would 
like to set a time table for addressing these issues.

The following VaR development/improvement projects were identified (in order 
of priority):

1. NG Correlations 
 Action steps:  - finalize and describe the alternative methodologies 
(Research, RAC)
   - evaluate the methodologies and select a superior one (Research, RAC, 
Traders)
   - implement the selected methodology (Research, IT)
   - test the results (Research, RAC)

2. Power Correlations (this methodology should be expanded to other 
commodities)
 Action steps: - implement an expanded correlation matrix for power 
(12-months) (Research, IT)
   - test the correlation matrix (Research, RAC)

3. FF Vols
 Action steps: - develop solutions for improving the existing methodology 
(Research)
   - implement the proposed solutions (Research, IT)
   - test the results (Research, RAC)

4. Jump Diffusion - NG
 Action steps: - implement the jump diffusion process for Gas (Research, IT)
   - test the results (Research, RAC)

5. Price Caps
 Action steps: - accumulate and summarize price cap information (RAC, Traders)
   - implement the price caps (Research, IT)
   - test the results (Research, RAC)

6. Intra-month Vols
 Action steps: - develop a methodology for smoothing the intra-month to 
monthly vol transition for power (Research)
   - implement the new methodology (Research, IT)
   - test the results (Research, RAC)
Vlady