Wednesday, September 29, 2010

An Interesting Discussion About Quant

I sometimes have the privilege of hearing from my colleagues in the money management community about the views expressed in my book, Buy and Hold is Dead (Again), as well as other articles that have been published in a variety of professional magazines and journals. The most recent correspondence I’ve enjoyed is with Brian Schreiner, Vice President of Schreiner Capital Management, Inc. Brian and I have exchanged views about the state of the money management industry in general, and more specifically the need for the industry to fully embrace the idea of active money management. Brian recently reread my book, and took issue with some of my ideas about quantitative decision making. For example, he correctly points out that in my book I call Modern Portfolio Theory the “father” (or something like it) of all quant models, when in fact, hardly anyone today thinks of MPT as being an example of quant. I feel that any technique for portfolio construction that doesn’t require reason, judgment, or informed intuition, but instead simply requires numbers to be plugged into a mathematical algorithm in order to asset allocate a portfolio, deserves the quant title. Here is what Brian had to say about quantitative portfolio management techniques:

"You get to my main point-- that investing is about making fewer mistakes. And, when it comes to mistake avoidance, computers are generally superior to human beings. Quant is the answer to the root problems of behavioral finance. The investor has to understand what they are-- an animal that has evolved into a mistake prone homo sapien, full of misperceptions, misunderstandings, bias, emotion, fears, greed, ego and so on… Working from there, you can start to develop an investment strategy. And, inevitably, you will arrive at a solution that relies not on the mistake prone human, but on a rule-based system that has one primary objective: to avoid mistakes.”

I thought Brian hit this one almost out of the park, and asked him for permission to share his thoughts with you in this blog. I agree with almost everything, except the part about quant leading you to a rules based system that avoids mistakes. It turns out that quant models are built by humans who suffer from all of the biases and heuristics that make qualitative decision making so difficult. The problem with a rules based system is that the rules change, and when they do someone has to recognize the changes and update the model. The process is excruciating. The model worked in the past but it hasn’t been working lately. Why? Maybe it’s the rise of high frequency trading, or hedge fund activity, or unanticipated or unprecedented government market intervention. Should we change the model? If so we are in new territory with no track record….and on and on. I still believe its best not to rely too heavily on quant as a matter of good risk management, the same way you need to defend against poor judgment calls. I believe that all investment models work…until they don’t. That’s why we use both quant models and informed intuition at Pinnacle. Thanks for letting me share your thoughts Brian. I always appreciate the comments of intelligent and passionate investors anxious to improve our industry and advance the cause of active portfolio management.