Hybrid Approach for Forecasting Stock Exchange Index Combining Statistical Methods and Artificial Neural Network

As stock exchange has a random and non-linear nature, predicting its related indexes with minimal error has been a major challenge for researchers and investors. In this paper, by reviewing the literature, the authors extracted used the representative indicators of different technical analysis and s...

Full description

Saved in:
Bibliographic Details
Published in:Optical memory & neural networks 2021-07, Vol.30 (3), p.194-205
Main Authors: Azad, Bavand Bayat, Pourrafiee, Mahdi, Amiri, Maghsoud, Azizi, S. Pourmohammad
Format: Article
Language:English
Subjects:
Complexity
Computer Science
Information Storage and Retrieval
Citations: Items that this one cites
Items that cite this one
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:As stock exchange has a random and non-linear nature, predicting its related indexes with minimal error has been a major challenge for researchers and investors. In this paper, by reviewing the literature, the authors extracted used the representative indicators of different technical analysis and statistical models to pre-process data. It was followed by selecting the best predictor variables for artificial neural network model, which predicts Tehran’s stock exchange index. The methods used to identify and select effective factors in the exchange index were stepwise regression and panel data models. Then, the multi-layer artificial neural network based on Levenberg–Marquardt and imperialist competitive algorithms were used to predict the intended index. Finally, the proposed model was tested on the top fifty companies in Tehran’s Stock Exchange and its efficiency was evaluated.
ISSN:1060-992X
1934-7898
DOI:10.3103/S1060992X21030073