Abstract
The authors present a simple data-driven decision support system for stock market trading using multiple technical indicators, decision trees, and genetic algorithms (GAs). It assembles technical indicators set into a decision tree based on stock trading rules and generates buy, hold, and sell classes that represent trading decisions. The main contribution of this study is the use of GAs based on a two-step classification method. This allows for selecting the relevant inputs and adapting them to the market dynamic. The GAs are used at the data input selection step and the weight selection step. Classifiers of different technical indicators are trained in the first step and combined into the trading rules in the second step. Random sampling and data input selection techniques were used to ensure the required variety of technical indicators in the first step. An evaluation shows that the proposed algorithm improved forecasting accuracy from 73.6% to 81.78%.