Predicting Systematic Risk Using Artificial Neural Networks (A Case Study: Saipa Company)

The purpose of this study is to introduce an artificial neural model for predicting systematic risk of Saipa Company by using macro economic variables. The net used in the study is a neural feedforward network with backpropagation algorithms. Price Index in Tehran Stock Exchange, Exchange Rate, Oil Price and Gold Price have been considered as four input variables, and the systematic risk as the output variable. The relevant data for each variable has been set weekly and monthly, and based on the data, 80 different neural networks were designed. The results show that the optimal model for predicting weekly systematic risk is a four- layer- model with the Root- Mean- Square (RMS) of 0/033314. Furthermore, the optimal model for predicting monthly systematic risk is a three- layer- model with the RMS of 0/065557.