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

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