Surprised their VAR (which is a crap way to measure risk) is not even halved in relation to before the crisis. People were definitely chucking it about back then, and the mood these days is like a morgue.
If you get better at measuring risk, VAR can go up. For instance, if your models assume that asset classes act in an uncorrelated manner, then VAR may be very low. You improve the model to capture correlation, and VAR for the same exact assets goes up.
Net - they could have a much less riskier position, but improvements in their risk modeling might not reflect it in VAR.
VAR can be good in conjunction with other metrics, but it's pretty awful as a standalone because it doesn't measure the severity of very rare tail events.
If you get better at measuring risk, VAR can go up. For instance, if your models assume that asset classes act in an uncorrelated manner, then VAR may be very low. You improve the model to capture correlation, and VAR for the same exact assets goes up.
Net - they could have a much less riskier position, but improvements in their risk modeling might not reflect it in VAR.
VAR can be good in conjunction with other metrics, but it's pretty awful as a standalone because it doesn't measure the severity of very rare tail events.