Wednesday, July 7, 2010

Algorithms/heurisitics - known/unknown risks

It should not be surprising that different decision problems exist once you breakdown the type of problems faced. There is a continuum of knowledge between what we know versus what we do not know. For risk management, this means that some risks are well-defined while other are not so easy to describe For the well-known, we have well defined pay-offs and well defined probabilities for those pay-off. For the unknown risks, the pay-offs are not well-known and the probabilities are not easily measurable.

The type of decision process to be used has to match the information that is available. Decision-makers have to map the process into the amount of information that is know about a particular problem. For those problems where the risks or information is well-know, an algorithm can be used. This can be as easy as following a recipe. When there is a higher degree of unknown risks, it is harder to use a well-defined decision process. Probabilities become more subjective. For these types of decisions, heuristics or rules of thumb may better serve the decision-maker.

For investments, more complex decisions require more discretion and adaptive behavior. Quantitative investing is better served for the knowable risks.

No comments: