As I wrote in one of my last blog posts, "naked strategies" are quite simply highly efficient, cost-effective methods of coming to a decision quickly. These methods help investors avoid many of the delays that are frequently part of the decision-making process, such as the delays that occur when human decision-makers apply their judgment.

Comparing naked portfolio managers to traditional portfolio managers, the latter often uses far more information than is used in the naked model. People assume that humans judges are more reliable and sometimes even more cost-effective in their investment decision-making because they use a lot more information. Yet ironically, naked strategies often outperform human managers despite using far less information.

There are many examples of this out side the area or portfolio managment. Orley Ashenfelter, a Princeton professor, developed a model for predicting the value of wine futures based on just two variables (rainfall and temperature). That model proved much more reliable than the decision-making of wine speculators, who had the same data available to them plus the benefit of using the old "swish and spit" method. Another example: The Goldman chest pain decision aide used to make triage decisions for patients uses just four variables, yet it made more reliable triage decisions than the cardiologists and emergency room doctors at Cook County Hospital in Chicago who had far more information. Greenblatt's Magic Formula, which I discuss in The Naked Portfolio Manager, uses just two inputs, yet has produced impressive results relative to market averages.

One would think that with more information, human judges would be able to out-predict models, but the evidence clearly indicates this is not the case. The problem that human judges have is that when you have lots of information, it's sometimes difficult to determine what data is extremely important and what is almost irrelevant. With naked decision-making strategies, on the other hand, the method's construction tells you what data is truly important.

Good models focus on only the most important criteria necessary for making the decision. Models with a huge number of factors often dilute the value of the most important variables by averaging them with inputs of lesser value.

Decision-Making Best Practice #20: If a model has a very large number of inputs, it usually means that the model's creator has not done a very good job of identifying the most important inputs. Be very skeptical of these models.