A Neural Network Based Object-Oriented Framework for Simulating Stock Market Trading Strategies

Investment theory has as a major tenet the concept of efficient markets. In an efficient market all information is fully reflected in the price of a stock. As such, there is no trading strategy based on known data that can earn an abnormal profit. Using price and volume data, this design and subsequent model produces a weak form test of the efficient markets hypothesis. First, an Object-Oriented design is presented as a framework for utilizing a probabilistic neural network (PNN) for the purpose of earning abnormal profits. Next, this design is implemented and subsequently used to simulate the process of buying and selling stocks. Insight is provided for the PNN parameters while giving values that produce the best empirical results. As detailed in this study, the PNN shows promise as the basis for developing trading strategies that may earn abnormal profits.