THE APPLICATION OF AGENT TECHNOLOGY IN STOCK PRICE PREDICTION

Abstract

Stock price prediction which is based on the movements of the stocks is used by investors to make the right decision (buy or sell) at the right time. Artificial Neural Networks (ANN) is one of the tools used for prediction. Accuracy of prediction depends heavily on the data learned by the ANN. Therefore the more data learned, the better the accuracy. This paper discusses the use of agents in ANN for stock price prediction. The architecture of the multiagent system and the tasks of the three agents are also discussed. The agent system was compared with the traditional ANN system based on the time required to complete all the processes, the cost involved and the effect of additional data on the accuracy performance of the ANN. It was found that the agent system recorded completion of all the processes earlier than the traditional ANN system. This consequently contributed to lesser cost. The accuracy performance of the ANN increased with additional data collected daily by the agents. Conclusively, the use of agents in stock price prediction reduced time and cost and increased the accuracy of the ANN’s prediction.