A hybrid approach for automated trading systems

Automated trading systems have been proved to be very helpful to traders and investors. However, most of the existing systems implemented with any statistical or neural network approaches are incapable to give accurate advice because they cannot capture the off-market information for which most of it is non-numeric. In this paper, an automated trading system which integrates the techniques from technical analysis, pattern recognition, and knowledge-based system is presented. The numeric data from the stock is processed by both the statistical method and feature extraction. The intermediate results are then used by the rule-based system which captures the off-market information in form of facts and rules. After that, the advice can be generated through the inference with the data stored in the database. For resolving the inconsistencies among the intermediate results from different components of the system, concepts from the fuzzy set theories are employed. Further, some data from the past Hang Seng Index (HSI) is used for illustration