Stock investment decisions by using genetic algorithms

A comprehensive review of the literature on financial research has shown that the use of artificial intelligence techniques has been steadily increasing in this field. The performance of the trading strategies, forecasting rules generated by these techniques has been evaluated on the data on many financial markets around the world. Although, the past decade has witnessed a flurry of interest within the financial industry and financial literature regarding artificial intelligence technologies, including neural networks, fuzzy logic and genetic algorithms, applications of these techniques to Turkish financial data have received minimal coverage in the literature.In an attempt to fill this gap in the literature, this thesis presents an application of genetic algorithms on the Istanbul Stock Exchange (ISE) data. Using ISE National 100 Index data and the data on the selected stocks traded on ISE and utilizing the methodology of previous research on well established financial markets, I attempt to show that investors (agents) who are constantly in search of models which are suitable for the markets that continuously change can design trading strategies generated by genetic algorithms and these strategies are more successful than individual technical indicators. In this thesis, based on the assumption that investors prefer strategies based on technical analysis indicators, I investigate decision systems which are capable of discovering profit opportunities in the market by analyzing the nonlinear relationships between the technical analysis indicators. The main objective of this thesis is to show that in this decision system a model that can anticipate profit opportunities by investigating the complex relationships between the prices can only be successful if it is continuously updated. Moreover, investors can use this model to analyze the system itself.The analysis of the performance of genetic decision trees based on the complex relationships between the technical analysis indicators can also be regarded as a test of weak form of market efficiency. The results of this study show that it is possible to predict profitable opportunities in the Turkish market using genetic algorithms. This finding casts doubt on the weak-form efficiency of Turkish Stock Market.