You can’t take the human out of quant models

Clearly you don’t write a blog entitled “What My Intuition Tells Me Now” and rely solely on quantitative methods.  As the retired co-portfolio manager of the Davis Appreciation and Income fund I learned very early in my career that facts would only take you so far in your investment process.

The problem is that facts, the thing that most investors spend their time analyzing, have all occurred in the past.  But investment returns unfold in the future.  Hmmm.  What to make then of investors who rely upon so-called quant models?  Generally the thinking is: computers can analyze data faster than humans and can therefore act faster than humans.

Not only that, but the thinking also goes something like: humans, with all of their emotional flaws, cannot act in the objective fashion that computers can.  Blah, blah, blah.  As I wrote back at the end of November, solely analytical methods are doomed to failure.  The reason is that you cannot escape the human-ness in the investment process.

I want to state upfront here that I believe in using technology as a tool to assist humans to make investment decisions.  I also agree that emotions are a tremendous detriment to economic and investment analysis.  But where I disagree is that the emphasis should be put on a computer to solve this ages-old investment problem.

As I write about extensively in my book The Intuitive Investor, I feel that to eliminate your emotions from the investment process you begin with the human, not run away from the human.

After all, who invented mathematics?  Humans.  Who discovered ways of utilizing the on-off properties of electricity?  People.  Who invented binary mathematics?  Yup, homo sapiens sapiens.  Who created the stock market?  Perhaps a tired refrain, but it was people.  And so forth.  You cannot take the human out of investing.

Which brings me to a story in the Wall Street Journal yesterday that is just like literally hundreds of other stories just like it: [cue: dramatic music] vaunted, immortal, not imperfect like humans, quant investment models have flaws!

AXA SA, a French insurance company, is paying a $25 million fine to the Securities and Exchange Commission and repaying $217 million back to aggrieved investors, because it hid from its clients a glitch in one of its quantitative investment models.  You see, a coding error disabled some risk-management controls.  The cost to AXA, $242 million, is not inconsequential.  You can’t remove the human from the investment equation.

Most of those who would disagree with me would focus on the fact that quant models, when executed without glitches, have none of the disadvantages of the human.  But to emphasize the human-ness of the investment business what I want to focus on briefly is what AXA management did afterward.

Warning: Those of you concerned with the lack of ethics on Wall Street or in some areas of capitalism should grab yourself an air sickness bag.

To do so I am going to generously quote the Wall Street Journal:

“After a junior employee discovered the error in June 2009, a “senior official” ordered staff to keep quiet and not fix the problem, the SEC said Thursday. The agency said clients who voiced “substantial concerns” about the poor performance of their portfolios were told it was due to market volatility and other factors, not the model’s failings.

“In October 2009, the unidentified senior official told an AXA Rosenberg board meeting convened to discuss the poor investment performance that he was “not aware of significant” mistakes in the model, the SEC said.”

Now that’s what I call ethical stewardship of shareholder monies!

So who was the “senior official?”  According to the Wall Street Journal’s sources it was none other than quant superstar himself, Barr Rosenberg.  For over 20 years Barr has been a proponent of quantitative methods for making lots of money.  And he has been very successful.

But that’s not the point.  The point is that if you want to become a better, less emotional investor then you have to begin with the human, not run away from him or her.  You can’t take the human out of quant models or out of the investment process.  A lesson that has cost the firm of one of the biggest quant geniuses $242 million to figure out.  Duh!

Jason