Abstract:
This research presents the genetic algorithm to build a trading system, which composes of a buying rule, selling rule and exit rule generated by combining technical indicators. The evolved trading systems are trained with historical data of SET50 starting from January 2001 to December 2010, and tested with the data from January 2011 to December 2015. The training result in terms of an average annual rate return is 20.06%, and the average annual return of the test data is 17.05%. In addition, the performance of the generated trading systems are compared with the MACD and the LARG using the data from SET50. The test uses historical data from January to December 2015. The proposed method yields 95.71% profit return. A profit return of while MACD and the LARG offer the return of 12.75% and 17.28% respectively. These show that proposed system can create a powerful investment strategy in terms of maximizing profit.