Technical Trading Rules in the European Monetary System

This paper describes the use of genetic programming to create trading rules based on past exchange rates and interest differentials. In an out-of-sample period, 1986-1996, these rules make significantly positive excess returns for three of four EMS exchange rates. There is evidence that the returns to the rules are higher when bands are wider, as a result of an increase in the variability of the exchange rate. There is no evidence that returns can be explained by conventional measures of risk. The rules outperform commonly used moving average and filter rules and do not appear to exploit previously known features of target zone exchange rates such as mean reversion