An Emedded Fuzzy Knowledge Base for Technical Analysis of Stocks

摘要

The techniques of technical analysis are commonly used in automate trading system, today. Many of the oftware tools have charting features with various technical indicators calculation, but human knowledge in reasoning is usually not incorporated and it is often left to the user for manual interpreaation. Although some types of decision aids are present, there is a lack of flexibility and general tools for handling uncertainties and human type of reasoning. Our development is an approach for technical analysis. comprising a flexible knowledge-based component embedded ain a general framework with graphic utilities and computation routines for technical indicators. The system has been developed under UNIX, with Motif as the graphic interface and FuzzyCLIPS as the knowledge component. The rules can be modified without the need of recompliation, and the user can observe interactively the consequences of machine interpretation using idfferent rule based. With simple heuristics to integrate knowledge of the different technical indicators, we obtaiend promising results for some pracitical cases. A reasoning scheme of combining evidences from the indicators based on Dempster-Shafer's Theory is also suggested.