Scholarlink

<p>Accurate forecasting of gold prices plays a crucial role in financial decision-making and risk management. However, modeling gold price dynamics remains challenging due to their nonlinear and temporal characteristics. In this study, we propose a multivariate deep learning framework based on a stacked Long Short-Term Memory (LSTM) network for short-term gold price forecasting. The model leverages multiple input features, including Open, High, Low, and Close (OHLC) prices, and employs a sliding window technique to capture temporal dependencies in time series data. Experimental results on real-world gold price data demonstrate that the proposed model achieves high predictive performance, with a Mean Absolute Percentage Error (MAPE) of 2.34% and a coefficient of determination (R²) of 0.9588. In addition, a comparative analysis shows that the LSTM model significantly outperforms the traditional Recurrent Neural Network (RNN) in terms of stability, accuracy, and robustness. Further analysis reveals that the model is particularly effective for short-term forecasting, maintaining low prediction errors across multiple time horizons. Despite slight performance degradation over longer horizons, the overall error remains within an acceptable range. These findings confirm the effectiveness of LSTM-based architectures for financial time series forecasting and highlight their potential for practical applications in investment analysis and decision support systems.</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 multivariate LSTM-based framework for accurate short-term gold price forecasting in Vietnam

  • 期刊论文

  • 发表日期：

    2026年8月3日

查看全文
  • Phat Nguyen Huu,
  • Bach Dang Pham,
  • Tu Tran Thi Thanh

摘要

Accurate forecasting of gold prices plays a crucial role in financial decision-making and risk management. However, modeling gold price dynamics remains challenging due to their nonlinear and temporal characteristics. In this study, we propose a multivariate deep learning framework based on a stacked Long Short-Term Memory (LSTM) network for short-term gold price forecasting. The model leverages multiple input features, including Open, High, Low, and Close (OHLC) prices, and employs a sliding window technique to capture temporal dependencies in time series data. Experimental results on real-world gold price data demonstrate that the proposed model achieves high predictive performance, with a Mean Absolute Percentage Error (MAPE) of 2.34% and a coefficient of determination (R²) of 0.9588. In addition, a comparative analysis shows that the LSTM model significantly outperforms the traditional Recurrent Neural Network (RNN) in terms of stability, accuracy, and robustness. Further analysis reveals that the model is particularly effective for short-term forecasting, maintaining low prediction errors across multiple time horizons. Despite slight performance degradation over longer horizons, the overall error remains within an acceptable range. These findings confirm the effectiveness of LSTM-based architectures for financial time series forecasting and highlight their potential for practical applications in investment analysis and decision support systems.