Random Forest Classification for Indonesian Blue-Chip Banking Stock Trend Prediction

Abstract

This study presents a Decision Support System (DSS) based on the Random Forest algorithm to predict stock price trend direction for Indonesian blue-chip banking stocks, specifically BBRI, BBCA, BMRI, and BBNI. The rapid development of fintech and digital investment platforms in Indonesia has intensified the need for intelligent, data-driven tools to assist retail investors in making informed decisions. Despite growing research on stock prediction, studies that simultaneously evaluate multiple Indonesian banking blue-chip stocks using ensemble machine learning methods remain limited. This study uses one year of historical Open-High-Low-Close-Volume (OHLCV) data from May 2025 to May 2026, comprising 234 clean observations per stock after feature engineering. Sixteen technical indicators were extracted as input features, including Moving Averages (MA5, MA10, MA20), Relative Strength Index (RSI), MACD, Volatility, and Volume Ratio. A binary classification label was assigned based on next-day closing price direction. The Random Forest model was trained on 80% of the data (187 samples) and tested on the remaining 20% (47 samples). Experimental results demonstrate that BBNI achieved the highest F1-score of 0.5965, while BBCA and BBNI both reached an accuracy of 51.06%. The study concludes that Volatility and MACD are the most influential features across all four stocks. This DSS framework provides a practical and reproducible foundation for intelligent investment decision support in the Indonesian banking sector.

GS Cited Analysis