|
|
Enron Mail |
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
|