An Intelligent Agent-Based Stock Market Decision Support System Using Fuzzy Logic

Fluctuations in prices and complexities of the stock market have made investment decision a difficult task. The paper proposes an Agent-based Fuzzy logic Stock Market (AFUSM) mode which incorporates agent technology, fuzzy logic, and technical analysis to address this problem. The architecture of AFUSM comprises three components: technical analysis, fuzzy inference, and mobile agents. In the technical analysis component, stock data captured from the Nigerian Stock Exchange were used to compute four chosen technical indicators. These are Moving Average Convergence/Divergence, Relative Strength Index, On-Balance-Volume and stochastic oscillator. These indicators served as inputs for the Fuzzy Inference System (FIS), which were expressed as fuzzy linguistic variables. The membership functions were defined as low, medium and high depending on the position of the chart indicator. The fuzzy IF-THEN rules involving the linguistic variables are a combination of the trading rules for each of these indicators. The output of the fuzzy inference is an investment recommendation of buy, sell or hold, which serves as input to the trading agents for automated trading. AFUSM was implemented in MATLAB 2013 and JADE 4.3. A total of 236 data points were captured from two Nigerian banks and used for testing the system. The results showed that the AFUSM agent made over 83% correct trading decisions. This showed that AFUSM could be deployed as an efficient decision support tool to aid investors in decision making.