Back Propagation ANN in Stock Market Forecasting
|Yasen Rajihy 1, Kesra Nermend 2, Akeel Alsakaa|
1. University of Babylon (UOBABYLON), Babylon 09332, Iraq
Stock market forecasting plays a key role in practice and theory of investment, especially taking into account the progress in automation of turnover on capital market. It is regarded as a difficult task in financial time series prediction. Traditional techniques such as statistical analysis, fundamental analysis and technical analysis are no longer valid in this field. Artificial Neural network (ANN) is a more beneficial technique for stock market forecasting than others. Back Propagation (BP) algorithm is one of the most popular neural network training algorithms for financial time series prediction. One of the biggest problems regarding the use of BP algorithm is the number of hidden layers and the number of neurons in each hidden layer. In this paper we present a best model of forecasting the stock market using different models of BP algorithm depending on the number of neurons in the hidden layer. WIG20 (Capitalization-Weighted Stock Market Index of the twenty largest companies on the Warsaw Stock Exchange) is used as an illustrative example to evaluate the performance of the proposed model. The experimental results show that the model of architecture consisting of an input layer with N neurons, one hidden layer with 3/2(N+1) neurons and output layer with one neuron outperforms the other models.
|Auxiliary resources (full texts, presentations, posters, etc.)|
Presentation: Oral at Current Economic and Social Topics 2015, by Yasen Rajihy
See On-line Journal of Current Economic and Social Topics 2015
Submitted: 2015-10-20 21:53 Revised: 2015-11-11 10:36