Designing a model for using artificial neural networks to predict nonlinear time series (Case study: Tehran Stock Exchange Index

Subject Areas : Futurology

Bahman Ashrafijoo ¹ , Nasser Fegh-hi Farahmand ^{2 *} , Yaghoub Alavi Matin ³ , Kamaleddin Rahmani ⁴

1 - Department of management, Tabriz Branch, Islamic Azad university, Tabriz, Iran
2 - Department of Management, Tabriz Branch, Islamic Azad University, Tabriz, Iran (Corresponding Author) [e-mail]
3 - Department of Management, Tabriz Branch, Islamic Azad University, Tabriz, Iran
4 - Department of Management, Tabriz Branch, Islamic Azad university, Tabriz, Iran

Received: 2022-05-14 Accepted : 2024-07-08 Published : 2024-08-01

Keywords: Total stock index, Predict, Artificial Neural Network, Tehran Stock Exchange,

Abstract :

Predicting the total stock index is a challenging task, due to the complexity of stock market variables and the lack of management and the occurrence of problems in critical situations, it is very difficult to develop an efficient model for predicting the total stock index. One of the important tools used for investment decisions is forecasting techniques, which are an integral part of the decision-making and control process. On the other hand, the accuracy of forecasting is directly related to decision risk. One of the well-known and new methods for predicting the total stock index is the method of using artificial neural networks. this research is applied in terms of purpose and descriptive based in terms of research method. It is analytical-mathematical in terms of survey and survey method. The statistical population of this research is the total index of TEPIX from 1369 to 1399. In this research, the tool that has been used to measure the desired variables is the documents and statistics of TEPIX and to analyze the data of this research, descriptive statistics and inferential statistics as well as multi-layer artificial neural network. Perceptron has been used. The results of this study show the confirmation of high accuracy of forecasting the total index of TEPIX compared to other estimation methods provided by the model, which has the power to predict the total index up to 1.7% error and also confirms the adherence to the index of TEPIX. A non-linear process is another result of this research.