Rule selection, fuzzy logic and time series

It is difficult to extract trading rules for financial trading systems. One reason is that the data often leads to conflicting rules, i.e., rules that have the same left-hand side but different right hand side. In this paper, we explore ways to deal with the problem of inconsistent rules using US ten-years bond time series data. We explore Wang's method, and identify a particular shortcoming. In addition, we explore the use of frequency, and the last-in-sequence methods in choosing among inconsistent rules. We conclude that frequency can be used to develop short-term systems, while last-in-sequence produces overly biased rules. >