A study of the parameters of a backpropagation stock price prediction model

Reports an empirical study of an artificial neural network which implements an experimental backpropagation stock price prediction model. A backpropagation neural net stock prediction model was constructed to test its prediction capability. The parameters were varied and the corresponding predictive results were recorded. The parameters studied in this research were the learning rate, momentum, number of neurons in the hidden layer, activation function and input noise. The artificial neural network model has been treated by many as a black box that takes inputs to produce a desired output. This research attempts to study the behavior of this black box when its parameters are altered.<<ETX>>