Scholarlink

<p>Investors and other equity market participants need accurate stock market forecasts to create effective investment plans. A prediction model's increased accuracy, even by small amounts, can result in significant financial gains. The intricacy and volatility of the stock data, however, make stock market prediction a challenging study subject. Deep learning algorithms have demonstrated potential in recent years in producing reliable predictions for sequential data. This study aims to improve stock market predictions for the Hang Seng Index (HSI) by addressing these problems by utilizing the historical data gathered from January 2, 2015, to June 29, 2023, introducing a novel model that integrates the Recurrent Neural Network (RNN) model with the Moth-Flame Optimizer (MFO), genetic algorithm (GA), and Ant Lion Optimizer (ALO). The results of this study demonstrate that the proposed hybrid ALO-RNN model performs more effectively than the single RNN model, employing evaluation metrics across different training data proportions. The test set demonstrates a significantly low root mean square error (RMSE) of 323.50 and high R-squared values of 0.9821. This improved accuracy lowers prediction errors, highlighting the durability and dependability of the model. The benchmark&#xa0;models—Transformer, Extreme Gradient Boosting, CatBoost, Long Short-Term Memory, and Cascade Long Short-Term Memory— were compared to confirm the proposed model's efficacy and resilience. It was also confirmed that the suggested model could generalize and make highly accurate predictions in other markets. Additionally, evaluations of the Apple, Dow Jones, and Nasdaq markets were conducted. Based on the accuracy of the proposed model, it was determined that the model ALO-RNN can be an advantageous tool for investors to lower investment risk and boost profit on investment.</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

Enhancing stock market precision estimate through the implementation of a hybrid methodology

  • 期刊论文

  • 发表日期：

    2026年5月4日

查看全文
  • Jia Liu,
  • Yuee Guo

摘要

Investors and other equity market participants need accurate stock market forecasts to create effective investment plans. A prediction model's increased accuracy, even by small amounts, can result in significant financial gains. The intricacy and volatility of the stock data, however, make stock market prediction a challenging study subject. Deep learning algorithms have demonstrated potential in recent years in producing reliable predictions for sequential data. This study aims to improve stock market predictions for the Hang Seng Index (HSI) by addressing these problems by utilizing the historical data gathered from January 2, 2015, to June 29, 2023, introducing a novel model that integrates the Recurrent Neural Network (RNN) model with the Moth-Flame Optimizer (MFO), genetic algorithm (GA), and Ant Lion Optimizer (ALO). The results of this study demonstrate that the proposed hybrid ALO-RNN model performs more effectively than the single RNN model, employing evaluation metrics across different training data proportions. The test set demonstrates a significantly low root mean square error (RMSE) of 323.50 and high R-squared values of 0.9821. This improved accuracy lowers prediction errors, highlighting the durability and dependability of the model. The benchmark models—Transformer, Extreme Gradient Boosting, CatBoost, Long Short-Term Memory, and Cascade Long Short-Term Memory— were compared to confirm the proposed model's efficacy and resilience. It was also confirmed that the suggested model could generalize and make highly accurate predictions in other markets. Additionally, evaluations of the Apple, Dow Jones, and Nasdaq markets were conducted. Based on the accuracy of the proposed model, it was determined that the model ALO-RNN can be an advantageous tool for investors to lower investment risk and boost profit on investment.