Application of Evolving Mamdani Fuzzy Inference System with Fuzzy Rule Interpolation and Extrapolation(eMFIS(FRI/E) for executing trades using quarterly earnings announcements

Fuzzy techniques have been studied for implementation in neural networks to better model the nature of stocks data in the financial markets. Various studies incorporating fuzzy techniques into neural networks have shown to increase interpretability and increase performance for modelling and predicting stock behaviour. One model adopting neuro-fuzzy techniques is proposed in a study by Susanti[10]. The model is a novel neuro-fuzzy system architecture called evolving Mamdani Fuzzy Inference System with Fuzzy Rule Interpolation or Extrapolation (eMFIS (FRI/E)) and it is studied for its application in predicting stock prices. Also, research has shown that there is a correlation between quarterly earnings reports and stock movements and that stock behaviour can be influenced by the dates of quarterly earnings report. The objective of this study is to propose a trading mechanism adopting the use of the eMFIS(FRI/E) architecture by Susanti[10] that will determine the time of trade execution and the predicted price swing given the knowledge of announced dates of earnings reports. The mechanism proposed is a 2-stage process that will determine the time of execution for the trade to be made and also the price swing from date of the execution of the trade to the date of event. The mechanism performance is assessed by its accuracy in predicting the price swing and the date for trade execution. The results of this study show that the proposed mechanism is able to predict time of trade execution to be made and the price swing, hence its application for decision making in executing trades is promising.Bachelor of Engineering (Computer Science