Thanks for the detailed and well-written explanation, appreciate it !!
I was finally able to work through everything carefully and now I get plausible VaR results.
The most important clue was that you pointed me to the specification of the sensitivities in sensitivity.xml.
Somehow I had changed the relative shift size for Delta from 1% to 0.1% and obviously this didn’t match my parameters in covariance.csv.
May I suggest that you include your answer in the official userguide ?
I am convinced that this explanation will be very helpful for other users as well to give more information about VaR calculation and covariance.csv.
By the way: I saw no product type for stocks in portfolio.xml, right ? At the moment, I use equity forwards as a workaround, but I am not sure if this is the best approach.