A stock price forecasting application using neural networks with multi-optimizer

Abstract

This paper proposes an application prototype for forecasting of stock prices using feed-forward neural network with back propagation, Particle Swarm Optimization and Differential Evolution. The prototype provides a convenient graphical user interface that allows choosing stocks, period of data, percentage of training set, technical indicators for model inputs and other algorithmic parameters. Multithreading is provided for efficient running and the downloaded historical data and forecasted output can be save for future use. An experiment was performed to investigate the performance of the three algorithms as well as the effects of number of hidden nodes of the neural networks.