Five things the August Quant meltdown Should Teach Investors.

The best available evidence and arguments suggest that the Quant Meltdown was caused by someone dumping a lot of the stocks quant traders liked and buying and buying a lot of the stocks quants didn’t like.  Quant theorists believe this person did so because he had to finance positions in a wholly different sector, sub prime loans, perhaps.  This theory may not be true (and click here to see my analysis of the theory) but the theory and the supporting evidence is certain enough to tell, at least, five things.

1. The Quant meltdown does not signal a quant apocalypse.  There are reasons to believe that certain quant approaches are over invested or close to it, but the meltdown isn’t one of them.  I wouldn’t divest from quantitative approaches on the basis of the August meltdown.
2. We can’t trust our statistical techniques. Amir E. Khandani and Andrew W. Lo, financial engineers at M.I.T and authors of a recent report on the meltdown state that the meltdown was a twelve standard deviation event.  Have you noticed how often we have been having ten and fifty standard deviation events these days?  You know how many of these you can expect to see during a lifetime?  None.  Something is radically wrong in the statistical techniques we are using. 
3.  We don’t understand our industries systemic risks.  Systemic risks are the risks you can not diversify away.  Assume the quant traders are right, what then is the link between quant long/short positions and sub-prime loans? As far as I can tell, there is no such link.  The link, if there was one, may have been the need of one investor to finance investments in one sector by closing positions in other sectors. This may bring comfort to the quants, but it makes your job and my job harder and uglier.  Systemic connections should be a little bit harder to find than investment profesionals on Wall Street, but not much harder.  But the kinds of connections we need to find are likely to be tenuous and hidden.  I see no reason why the investor couldn’t raise money by selling gold, for example.    
4. We need to at least two separate types of techniques to manage our portfolios.  I have long been an advocate of robust statistical techniques.  These are techniques that assume that the most extreme observations are most likely wrong.  Most of the time this is exactly the right thing to assume.  But basing portfolio management system on robust measures is a prescription for disaster when things go bad, when we have twelve standard deviation events and worse.  We need another set of techniques that make exactly the opposite assumptions.  We don’t have those techniques right now.  The closest we have is extreme value theory and stress testing and those techniques are only close in the sense that New York, NY is relatively close to San Francisco as opposed to Pluto.
5. We need to keep more of our wealth in cash than our models suggest.  Cash is what you invest in when you don’t know what else to do.  And the available evidence shows we know less than we think.