Dollars and Jens
Thursday, February 05, 2004
Modeling Risk
I had class Tuesday night. It was a "risk management" class; I'm still not sure quite what the aim of that class is, whether it will emphasize the quantitative or the qualitative. But there was an important point made that I thought I'd pass along:

If your model says that some risk has a ten-to-the-minus-thirteenth probability, the probability to start worrying about is the probability that your model is flawed. This was the real problem, for example, at Long Term Capital Management -- they had too much faith in their models, and believed that a 2σ event was a 12σ event.

Powered by Blogger