Passed along by a colleague. A short article and podcast that makes a very profound point. No matter how sophisticated and mathematical a model is, if it does not contain all of the important context of problem, it be very risky indeed to use predictively. Deciding exactly what to include and leave out of a model is the most important part of modeling. Worst of all, people often automatically validate models that agree with their preconceptions. Its not that I do not believe in quantitative models, I practiced building them for years. It is that there has to be a balance between common sense and quantitative infrastructure. One way to embrace this is to use the technique of scenarios.
The risks of risk management
Banks now employ thousands of highly-qualified mathematicians to quantify risk for them. So why did they not foresee the credit crunch? Quantitative finance lecturer Paul Wilmott explains how a failure to see beyond the numbers might be to blame.... "
No comments:
Post a Comment