Integrating singular spectrum analysis and nonlinear autoregressive neural network for stock price forecasting

Asmaa Y. Fathi, Ihab A. El-Khodary, Muhammad Saafan

Abstract


The main objective of stock market investors is to maximize their gains. As a result, stock price forecasting has not lost interest in recent decades. Nevertheless, stock prices are influenced by news, rumor, and various economic factors. Moreover, the characteristics of specific stock markets can differ significantly between countries and regions, based on size, liquidity, and regulations. Accordingly, it is difficult to predict stock prices that are volatile and noisy. This paper presents a hybrid model combining singular spectrum analysis (SSA) and nonlinear autoregressive neural network (NARNN) to forecast close prices of stocks. The model starts by applying the SSA to decompose the price series into various components. Each component is then used to train a NARNN for future price forecasting. In comparison to the autoregressive integrated moving average (ARIMA) and NARNN models, the SSA-NARNN model performs better, demonstrating the effectiveness of SSA in extracting hidden information and reducing the noise of price series.

Keywords


data preprocessing; nonlinear autoregressive neural network; singular spectrum analysis; stock market; stock price prediction;

Full Text:

PDF


DOI: http://doi.org/10.11591/ijai.v11.i3.pp851-858

Refbacks

  • There are currently no refbacks.


View IJAI Stats

Creative Commons License
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.