Title: Verification of neural network models for forecasting the volatility of the WIG20 index rates of return during the COVID-19 pandemic

Authors: Emilia Fraszka-Sobczyk; Aleksandra Zakrzewska

Addresses: Faculty of Economics and Sociology, University of Lodz, ul. Rewolucji 1905 r. nr 37/39, 90-214 Lodz, Poland ' Faculty of Mathematics and Computer Science, University of Lodz, ul. Banacha 22, 90-238 Lodz, Poland

Abstract: The paper investigates the issue of the volatility of stock index returns on the Warsaw Stock Exchange (the WIG20 index returns volatility). The purpose of this review is to present an alternative neural network and to examine it to predict stock index returns according to the historical data. Finally, these predictions from the new neural network are compared with predictions based on a standard neural network MLP. In this article, the measurements for the best forecasting performance of neural networks are taken from commonly used forecast error measurements: mean error (ME), mean percentage error (MPE), mean absolute error (MAE), mean absolute percentage error (MAPE), root mean square error (RMSE), R (the correlation coefficient). The results show that the introduced neural network has good accuracy in measuring effectively the WIG20 index returns volatility.

Keywords: stock index returns? volatility forecasting? stock index prediction? neural network? machine learning.

DOI: 10.1504/IJADS.2024.137001

International Journal of Applied Decision Sciences, 2024 Vol.17 No.2, pp.137 - 155

Received: 17 Mar 2022
Accepted: 31 Aug 2022
Published online: 01 Mar 2024 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article