Unfortunate cosT Of Pattern rEcognition. This is the cover story of the Wilmott Magazine, Jul-12 issue. Apophenia is the technical name for a problem with human nature: seeing meaningful patterns, when they do not exist.

It is criticized that in finance such recognized but non-existing patterns are often used to convince investors or recommend investment strategies. One of the questions: knowingly or not? Apophenia or immoral behavior (tell appealing stories to get a sale)? In many cases, apophenia can fool both financial professionals and clients. Apophenia is a natural part of human beings' survival, but in the financial world it is problematic.
Often it is just a matter oft training to understanding the asymmetry in pattern recognition and  its consequences.
This is another argument against uncompromisingly managing personal invetsments or blindly rely on financial advisors?!

A jump to a more general discussion - we often hear:
  1. "Quant finance will never discover scientific laws, because human behavior changes"
  2. "Finance should not use math, because it can't describe human behavior"
  3. "Quant finance is not a science"
Even more for economics. But the theories that won the Economics Nobel Price this week, discover "laws" and "difficult" math is essential to extract them. Matching Theory (basically an algorithm).

I am not en expert in this theory at all, but I understand it is testable, tested and correct.

2009 I wrote Beware Wind - apophenia pure. And I agree, extracting knowledge from data is not easy. You need multi-strategy and multi-method approaches and even better combine math and machine learning methods intelligently. And pattern matching is easier than pattern recognition - and also it makes a difference if you search for patterns in input, system behavior or output.

In risk management it is often the analytics of risk spectra resulting from comprehensive portfolio-across-scenario simulation. This needs sophisticated models and mathematical schemes that might not describe the real world perfectly, but that are useful when calibrated adequately ....