Comparative Analysis Of Data Mining Methods For Indonesian Stock Prediction: Support Vector Machine (SVM) Vs Random Forest (RF)

The research presented in this paper examines the SVM and RF models for predicting BBCA.JK's market downturns between July 2022 and June 2024. BBCA.JK was classified as one of the Indonesian blue chip stocks. This period was when BBCA.JK had entered the four market regimes: normal, trending, volatile, and volatile trend. The study demonstrates the effectiveness of machine learning in market prediction, with the SVM model achieving 96% accuracy in downturn detection. Analysis of volatility thresholds between 25% and 30% proved optimal for identifying market downturns across different trading conditions, particularly during periods of under and overselling. BBCA.JK's downturn in prediction and detection using RF model in its total of 92 percent accuracy was tested during stable market conditions. The benchmarks for market correction were derived from a control function embedded in both models and used standard market volatility indicators for calibration. These results will contribute significantly to the trading strategies in working situations within less developed markets where emerging technologies specifically machine learning or AI can be embedded.