Forecasting stock prices using neural network and squirrel search algorithm

Document Type : Original Article

Authors

Homa Mehtarizadeh ¹

Najme Mansouri ²

Seyed Mohammad Mehdi Hosseini ¹

Razieh Delpasand ³

¹ Applied Mathematics, Shahid Bahonar University of Kerman, Kerman, Iran

² Faculty of Shahid Bahonar University of Kerman

³ -PHD of Operations Research, Shahid Bahonar University of Kerman, Kerman, Iran

https://doi.org/10.22067/mfe.2024.75274.1167

Abstract

‎Investing in stocks offered on the stock exchange is one of the most lucrative investments in the financial and capital markets‏. However, stock price behavior is one of the most complex mechanisms that researchers have studied. This market always shows unpredictable, chaotic, and non-linear behavior because it is affected by economic, political, and industrial conditions. In this paper, we try to predict the closing price of the share in the coming days by using Multilayer Perceptron Neural Networks and Elman. The weights of the neural networks used in this work are optimized using the squirrel search algorithm, which increases the speed and accuracy of the neural networks. The results are evaluated by MSE, RMSE, MAE, and MAPE criteria. Also, the proposed method is compared with Particle Swarm Optimization (PSO), Genetic Algorithm (GA), Grey Wolf Optimization (GWO), and Artificial Bee Colony (ABC) metaheuristic algorithms to show that the optimization of network weights using squirrel search algorithm averages 72% for Iran Telecommunication Companies, National Iranian Copper Industries, Pars International Products, Ghadir Holding and India Steel has better results than other methods. In comparison with the multilayer perceptron neural network with Elman neural network, Elman neural network has an average of 27% better results than the multilayer perceptron neural network.

Keywords

Artificial neural networks

Forecasting

Meta-heuristic algorithms

Squirrel Search Algorithm

Subjects

• take stock

---