It's mild evidence against the "top level math talent will be able to identify bubbles before they burst" hypothesis, but only mild because I don't think Madoff was transparent enough to be obviously unstable in the way that the housing market was.