Artificial intelligence has finally hit its stride with the growing popularity of rule-based expert systems. The technology is already being applied in medicine and the sciences, and its application to such investment tasks as asset allocation and performance attribution is just around the corner. Will expert systems represent a revolution or an evolution for investment management?

In investing, the computer has been applied extensively and profitably in simulation and screening. Expert systems are similar to these basic approaches, but go well beyond them in range of application, ease of use and flexibility. An expert system can indicate all possible solutions to the goal set for it, describe how those solutions were found and give a quantitative measure of how likely it is that each solution is true.

The key to an expert system is its knowledgebase—the set of rules the system applies to the database of facts. The output from an expert system will be only as good as the rules laid down for it. Construction of a viable system requires both experts in the knowledge required to develop the rules and expertise in molding that knowledge into a reasonable number of properly formulated rules. It is important to work from the outset toward a goal that is neither too broad nor too narrow, allowing enough time for design and testing of the system.