Scholarlink

<p>This study presents a new hybrid forecasting system to enhance the accuracy and efficiency of predicting stock market trends. To do this, the proposed involves several steps. Firstly, the closed stock index price time series is preprocessed to address missing values, variance stabilization, nonnormality, and nonstationarity. Second, the stock index closing prices are processed and filtered into a nonlinear long-term trend series and a stochastic series using three proposed filters and a benchmark filter. Third, the filtered series are estimated using the nonlinear and neural network autoregressive models. Fourth, the residual from both the fitted models is extracted, and a new series is obtained. The new residual series is forecasted using the autoregressive condition heterogeneity model. Finally, the forecasts from each model are combined to get the final estimates. The results indicate that the proposed final hybrid model produced the most accurate and efficient comparison with the baseline models.</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

A novel hybrid framework for forecasting stock indices based on the nonlinear time series models

  • 期刊论文

  • 发表日期：

    2025年3月11日

查看全文
  • Hasnain Iftikhar,
  • Faridoon Khan,
  • Elías A. Torres Armas,
  • Paulo Canas Rodrigues,
  • Javier Linkolk López-Gonzales

摘要

This study presents a new hybrid forecasting system to enhance the accuracy and efficiency of predicting stock market trends. To do this, the proposed involves several steps. Firstly, the closed stock index price time series is preprocessed to address missing values, variance stabilization, nonnormality, and nonstationarity. Second, the stock index closing prices are processed and filtered into a nonlinear long-term trend series and a stochastic series using three proposed filters and a benchmark filter. Third, the filtered series are estimated using the nonlinear and neural network autoregressive models. Fourth, the residual from both the fitted models is extracted, and a new series is obtained. The new residual series is forecasted using the autoregressive condition heterogeneity model. Finally, the forecasts from each model are combined to get the final estimates. The results indicate that the proposed final hybrid model produced the most accurate and efficient comparison with the baseline models.