A way of VaR attribution

For me, one of the most frequently asked question from the client is, “What caused the change in VaR numbers?” When asked, I just skim over daily position change report, pick up a couple of notable changes, calculate rough VaR change using position value and risk weights, and when the number roughly matches the difference in VaR numbers, then it’s done. However, when there are several hundred position entries (commodity, location and delivery period), it’s difficult to carry out the procedure. Moreover, many of the notable changes are just ‘rebalancing’ trades – maturity roll over, liquidate overpriced and purchase underpriced instruments, replace swaps with options, and so on. These trades have marginal VaR impact when coupled with their offsetting positions, wasting your time for analysis.   So I was thinking of a smart or at least more automated way of detecting ‘notable’ position changes that have substantial VaR impact  Continue Reading in LinkedIn