Reliable stock price prediction is critical for effective investment decisions; however, high volatility and nonlinear dynamics continue to challenge forecasting accuracy. Despite the extensive use of machine learning in financial research, short-term comparative studies on Indonesian banking stocks remain scarce. This study evaluates the performance of Support Vector Machine (SVM), Long Short-Term Memory (LSTM), and Random Forest models in predicting Bank Mandiri’s stock prices using daily data from Yahoo Finance covering June to December 2024. The data, including price indicators and trading volume, were normalized, transformed into time-series sequences, and divided into training and testing sets. SVM was applied for directional classification, while LSTM and Random Forest were used for regression-based price prediction. Model performance was assessed using accuracy and mean squared error (MSE). The findings show that LSTM achieves the lowest prediction error (MSE = 0.0045), indicating superior ability to model temporal and nonlinear price patterns. In contrast, Random Forest records the highest classification accuracy (0.9932), demonstrating strong performance in predicting price direction. Overall, LSTM is most effective for short-term price forecasting under volatile market conditions, whereas Random Forest remains a robust option for directional classification.
Rambe et al. (Wed,) studied this question.