Scholarlink

<p>The noisy, non-stationary, and nonlinear nature of financial time series data is a critical issue in predicting stock prices accurately. Improved decision-making in the financial markets and reducing the risks of investments require the development of efficient forecasting models. Concluding the historical data obtained from the KOSPI between January 2018 and January 2024, the current study tackles the enormous challenge of predicting stock prices. Various decomposition approaches were employed because financial time series are nonlinear. Variational Mode Decomposition (VMD) is selected to adaptively decompose intrinsic mode functions (IMFs) that are more effective in extracting the multiscale features of the original signals. Artificial Ecosystem-Based Optimization (AEO) is added as a metaheuristic search algorithm to guarantee the optimal setting of model parameters and improve generalization. The Radial Basis Function (RBF) network, the main predictive element, is especially suitable for financial forecasting applications due to its high nonlinear function approximation capability, quick convergence, and low computational cost. The suggested framework consistently outperforms the models because of its superior accuracy and stability. Its superiority in revealing nonlinear interdependencies and dynamic patterns in finance time series is attested to by its <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(\:{R}^{2}\)</EquationSource> </InlineEquation> of 0.9955 on the KOSPI dataset. Generalizability of the proposed model was confirmed by applying it to a variety of financial markets, such as the Dow Jones Industrial Average, gold, and crude oil, and achieving substantial <InlineEquation ID="IEq2"> <EquationSource Format="TEX">\(\:{R}^{2}\)</EquationSource> </InlineEquation> values of 0.9941, 0.9927, and 0.9925, respectively. These findings demonstrate how adaptable and robust the model is to a variety of market conditions.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Stock Price Prediction With Variational Mode Decomposition, Ecosystem-Based Optimization, and Radial Basis Function Models: Korea Composite Index Insights

  • 期刊论文

  • 发表日期：

    2025年10月22日

查看全文
  • Yun Yang,
  • Hong Liu,
  • Shaoming Yang

摘要

The noisy, non-stationary, and nonlinear nature of financial time series data is a critical issue in predicting stock prices accurately. Improved decision-making in the financial markets and reducing the risks of investments require the development of efficient forecasting models. Concluding the historical data obtained from the KOSPI between January 2018 and January 2024, the current study tackles the enormous challenge of predicting stock prices. Various decomposition approaches were employed because financial time series are nonlinear. Variational Mode Decomposition (VMD) is selected to adaptively decompose intrinsic mode functions (IMFs) that are more effective in extracting the multiscale features of the original signals. Artificial Ecosystem-Based Optimization (AEO) is added as a metaheuristic search algorithm to guarantee the optimal setting of model parameters and improve generalization. The Radial Basis Function (RBF) network, the main predictive element, is especially suitable for financial forecasting applications due to its high nonlinear function approximation capability, quick convergence, and low computational cost. The suggested framework consistently outperforms the models because of its superior accuracy and stability. Its superiority in revealing nonlinear interdependencies and dynamic patterns in finance time series is attested to by its \(\:{R}^{2}\) of 0.9955 on the KOSPI dataset. Generalizability of the proposed model was confirmed by applying it to a variety of financial markets, such as the Dow Jones Industrial Average, gold, and crude oil, and achieving substantial \(\:{R}^{2}\) values of 0.9941, 0.9927, and 0.9925, respectively. These findings demonstrate how adaptable and robust the model is to a variety of market conditions.