عنوان مقاله [English]
One of the new areas in financial researches is using artificial intelligence to assist building decision making systems. Stock trading system is one of those systems developed to help investors make successful trading operations. Successful trading operation must be done near turning point of price trends. In recent years many studies focused on preparing systems to suggest price trend reversal. Technical analysis which tries to provide trading signals is mostly used in such systems and is usually one part of the system.
Technical analysis with a lot of rules try to give trader the signals of price trend reversal but the disadvantage of technical analysis is its dependency to investors experience to decide on technical rules and parameters. In fact the performance of technical analysis is deeply dependent to quality of setting technical parameters.
In this study we try to build a trading system based on technical rules and enhance its performance by using Genetic Algorithms, Fuzzy Logic and Artificial Neural Networks. GA helps us to train technical parameters in technical rules. Fuzzy logic helps us to discern how is the condition of market (trending market or none trending market). Because it is important to select kind of rules. When different enhanced rule provide their trading signal concerning market condition, an ELMAN network combines different signals together to provide trading suggestion.
Results from Tehran stock exchange consist of 10 stocks demonstrate that statistically there is significant difference between performance of our proposed system and grand trading strategies such as buy and hold strategy. In other words, our system possesses profitability potentials.