<![CDATA[Gold is a precious metal with high value that is often used as an investment commodity due to its stability and tendency to increase in price compared to other assets, such as stocks. In the global economy, gold is also an important part of international reserves in national banks. However, public awareness of the benefits of gold investment remains low. One solution to increase interest and understanding of gold investment is to predict gold prices using accurate forecasting techniques. Forecasting utilizes historical data that is analyzed to project future trends, making it an important component in strategic decision-making. This study uses the backpropagation algorithm in artificial neural networks to predict gold prices. This algorithm minimizes errors in the data training process, improves model accuracy, and provides better results in prediction classification. Additionally, this algorithm is efficient in processing large amounts of training data, resulting in a reliable prediction model. The study aims to evaluate the performance of the backpropagation algorithm in predicting gold prices, including comparing the accuracy and correlation of predictions with other algorithms. The results of the study are expected to contribute to the development of a more accurate gold price prediction model, support investment decision-making, and increase public understanding of the benefits of investing in gold. This study successfully developed an Artificial Neural Network (ANN) model to predict gold futures prices based on historical data, including features such as opening price, high, low, and trading volume. The model was trained using the Backpropagation algorithm to capture non-linear patterns in complex data. The research results encompass three main aspects: Data Preprocessing, where data was effectively processed, including converting values to numerical format and normalizing features to accelerate model convergence; Model Training, where the model was trained using 80% of the training data and tested with 20% of the testing data; Monitoring train loss and validation loss shows that the model is learning well, although there are indications of overfitting risk. Evaluation and Prediction: The model is able to predict gold prices with good accuracy on the test data. Evaluation metrics such as MAE (Mean Absolute Error) show that the prediction results are quite close to the actual values, although there is still room for improvement. Overall, this model demonstrates satisfactory performance in predicting short-term gold prices and can be used as a tool in gold price analysis based on historical data.]]>
