A Business Shuts its doors
It is always somewhat sad to hear the news of a business closing due to tough circumstances. Yesterday we heard from a good friend and colleague in the investment advisory world that such was the case for him.
This fellow has not only been a friend, he has been a mentor of sorts and was ahead of the technology curve in our industry by designing and maintaining his own portfolio testing and tracking software when others were using simple desktop applications that did a good job of managing phone numbers but not of bringing together the many facets of client needs, portfolio construction, and manager functions. Yet, we find ourselves wondering if the trail he followed (and which is followed by many) is simply a trail that inevitably leads if not over a cliff, then to a dead-end. At that trail-head is the sign “Mechanical Systems”. It should be underscored by the phrase “Follow at your own peril”.
Having spent a few years working at a research firm which tracks many managers who employ mechanical systems to manage trade decisions, we can say with some certainty that no mechanical system works all the time. Of course, the meaning of “works” is a function of what the system is designed to accomplish. We are not in agreement with those who say that beating market indices is the measure of success. Tell that to the folks we know who “only” lost 30% in ‘02-’03 when the S&P lost 40%. Rather, we think the measure of success is whether or not a trading system achieves the personal financial objectives of the individual investor over time. Therefore, by definition a mechanical system must have a one-size-fits-all philosophy with slight variations to returns and risk accomplished only through allocations to cash and equivalents. This makes whether a system works or not a question of whether or not it consistently outperforms some index over some arbitrary period of time.
Mechanical systems must eventually fail because the data on which their rules are founded is old data. It has to be! Future data is not yet available (except to crystal ball owners - but that is a topic for another day). Some systems try to fix this problem by constantly updating and testing and reforming the rules by frequent testing and optimizing of parameters for trade decisions. It is our view that this is not a bad thing, but it begs the question of whether or not backtesting is a truly viable means of making a decision about the future functionality of the parameters. And, the more often a system is re-optimized the less time it has history using a certain set of rules. Bringing us back to the point that re-optimization is necessary only because of the basic underlying pre-supposition that the system is going to fail unless it is re-optimized. And, unless the re-casting of rules is also mechanized, the system by default becomes other than mechanical once the decision to do so is not simply part of a pre-determined program.
There is no perfect system out there, but trading strategies can be formulated to take into account what is the worst that can happen if the trade decision is a wrong one. In our view the human brain is best equipped (albeit some are better equipped than others and we are making no claims to the quality of equipping in our own) to make the qualitative decisions that must be part of all trading on the macro and micro level. Formulas and computer models must be the tools the decision maker uses and not the decision maker themselves.
Our friend will bounce back, we are certain. We pause not to mourn but to remove our hat in honor of the things he did that helped to set some industry standards and set some industry trends. We also reflect on the pursuit of the mechanical holy grail which led, in part, to the demise of his firm. We will try not to make the same mistake.
