AI Networking Cookbook: Practical recipes for AI-assisted network automation and development
TL;DR
A multivariate deep learning framework based on a stacked Long Short-Term Memory (LSTM) network for short-term gold price forecasting that significantly outperforms the traditional Recurrent Neural Network (RNN) in terms of stability, accuracy, and robustness.
Accurate short-term gold price forecasting is crucial for the financial decision-making. This paper introduces a short-term gold prediction network (STGP-Net), a novel hybrid deep learning model designed to enhance prediction accuracy by integrating one-dimensional convolutional neural network (1D-CNN) and long short-term memory (LSTM). STGP-Net leverages the 1D-CNN's ability to extract local temporal features and the LSTM capacity to model long-range dependencies within gold price time series. Various sliding window configurations are employed to generate input sequences for multi-step ahead prediction. Comprehensive experiments were conducted comparing STGP-Net against 1D-CNN + recurrent neural network (RNN) and 1D-CNN + bidirectional long short-term memory (BiLSTM) baseline models across three configurations using metrics like mean absolute error (MAE), root mean square error (RMSE), and determination (R²). The results demonstrated that STGP-Net consistently provided better performance and robustness, proving more effective for short-term gold price forecasting than the alternative hybrid models tested.
Gold is viewed as an investment that will remain valuable over the long term and as an investment that will hedge against inflation; however, the volatility of its price in the short term necessitates the use of effective forecasting techniques for investment decisions. This research uses a Hybrid Deep Learning technique, by predicting the price of gold using historical time series data with a Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) model. The model was tested with batch sizes of 16, 32, and 64 using the Adam optimizer with a learning rate of 0.0001 and dropout of 0.2. This research provides an indication of the extent to which gold price forecasting, at least from a financial forecasting perspective, can be achieved using a hybrid model of CNN and LSTM, as it showed the capability to detect short-term trends and long-term sequential gaps in gold price series. The experimental results show out of several performed analyses on the CNN-LSTM model, the one with a batch of 16 showed the best performance as it achieved an RMSE (Root Mean Square Error) of 11.518535% which implies there was great closeness between the actual gold price and the predicted gold price
Inflation forecasting remains a complex problem due to nonlinear dynamics and interactions among macroeconomic variables, particularly in emerging economies such as Indonesia. Previous studies using deep learning models, including Long Short-Term Memory (LSTM), have shown promising results but often suffer from high variance and limited robustness, especially when temporal dependencies are not properly preserved. This study aims to develop a more stable and accurate forecasting model by integrating Bagging with a Stacked LSTM architecture using the Moving Block Bootstrap (MBB) method. The proposed model utilizes multivariate time series data consisting of inflation, exchange rate (USD/IDR), BI interest rate, and money supply, with preprocessing techniques including Z-score normalization and sliding window transformation. Experimental results show that the model achieves an RMSE of 0.4273 and MAE of 0.3048, indicating good predictive performance. Compared to baseline models such as ARIMA and single LSTM, the proposed approach provides more stable and consistent forecasting results. The model is also able to generate reliable predictions for the next 12 periods, demonstrating its ability to capture temporal patterns effectively. These findings suggest that the integration of Bagging, Stacked LSTM, and MBB improves model robustness and forecasting accuracy. The proposed approach can support data-driven decision-making in economic policy, although further research is needed to incorporate additional variables and explore more advanced forecasting architectures.
Stock price forecasting has become an important research area in financial market analysis due to the highly dynamic and volatile nature of stock market movements. Accurate forecasting models can assist investors and financial analysts in making better investment decisions and managing financial risks more effectively. This study proposes a Bidirectional Long Short-Term Memory (BiLSTM)-based deep learning model for forecasting stock prices using multivariate financial time-series data. The dataset used in this study consists of historical stock market data combined with several technical indicators, including High, Low, Open, Volume, Moving Average (MA10 and MA20), and Relative Strength Index (RSI). Prior to model training, the data underwent preprocessing stages including data cleaning, normalization, and sequence generation using a sliding window approach. The proposed architecture employs a Bidirectional LSTM layer followed by additional LSTM and dense layers to capture complex temporal dependencies and nonlinear patterns in stock price movements. The experimental results demonstrate that the proposed model achieved moderate forecasting performance with an MAE value of 15.5805, RMSE of 21.0883, and R² score of 0.4954. Furthermore, the prediction results show that the model was able to follow the general trend of actual stock price movements despite several deviations during periods of high market volatility. The findings indicate that the BiLSTM architecture has strong potential for financial time-series forecasting and can effectively capture important sequential patterns in stock market data. This study also highlights the challenges of stock price prediction due to the stochastic and nonlinear behavior of financial markets. Future research is recommended to incorporate additional external factors and more advanced deep learning architectures to further improve forecasting accuracy and robustness.
Stock price forecasting is a key part of investment decision-making, especially for high-capitalization stocks such as PT Bank Central Asia Tbk. (BBCA). Accurate prediction remains challenging because complex market dynamics and nonlinear price movements influence stock prices. This study proposes a multivariate Long Short-Term Memory (LSTM) model that integrates the closing price and the RSI-14 technical indicator as input features to improve accuracy over conventional univariate approaches. We obtained historical data from the Yahoo Finance API covering January 2015 to December 2025, totaling 2,698 trading days after RSI-14 feature engineering. The dataset was split to 80:20 for training and testing, with MinMaxScaler normalization applied only to the training data to prevent data leakage. The LSTM model used 100 neurons with an input shape of (10, 2) and was trained using the Adam optimizer with early stopping at epoch 24 to avoid overfitting. Evaluation results show an RMSE of 169.72 IDR, an MAE of 134.68 IDR, a MAPE of 1.58%, and an R² of 0.9368, indicating a good regression-level fit. Adding the RSI-14 feature improved the regression metrics relative to a univariate closing-price-only LSTM. Given the modest trend-classification accuracy and the lack of trading back testing, transaction costs, or risk-adjusted performance measures, view the results as a methodological contribution rather than direct evidence of practical investment value. Future work could extend this research by exploring hybrid architectures that combine LSTM with attention mechanisms or Transformer-based models, and by validating the approach on other high-capitalization stocks to assess generalizability across market conditions.
The results indicate that self-supervised representation learning can serve as an effective and scalable substitute for manual feature engineering in finance time-series forecasting.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.