THE APPLICABILITY OF ARTIFICIAL NEURAL NETWORK METHOD UPON PREDICTION OF RATE OF STOCK RETURN: EXAMPLE OF 2008 FINANCIAL CRISIS
THE APPLICABILITY OF ARTIFICIAL NEURAL NETWORK METHOD UPON PREDICTION OF RATE OF STOCK RETURN: EXAMPLE OF 2008 FINANCIAL CRISIS

Summary/Abstract: Main target of this study is to set a model that will make predictions based upon financial statements and companies’ rate of stock return in crisis periods when economic indicators are at high level and show sudden changes and to develop an artificial neural network model that has forecasting accuracy just as much as statistical models while setting this model. For this purpose, 20-period 400 data of 20 companies’ 2006/Q1-2010/Q4 periods that carry on their services in food sector and have been treated at Istanbul Stock Exchange (ISE) in periods 2006-2010 in Turkey were calculated. Artificial Neural Network (ANN) that was set in our study has 20-neuronal single hidden layer 21 input variables and one output variable. The model that we used is Multi Layer Perceptron Back Propagation Artificial Neural Network model. In our study, Mean Square Error (MSE) was determined as performance measurement with which training will be completed on achieving. With this model, Training Phase MSE value was found as 0.0309 and testing phase MSE value was 0.0502. It was determined that the developed multilayer perceptron model has had the capacity of forecasting 2008 crisis period successfully.

• Details
• Contents