Prediksi Dinamika Harga Indeks Saham Menggunakan SVM (Support Vector Machine) Dengan Pendekatan Time Series Moving Avarage.

Abstract

Penelitian ini mengevaluasi prediksi harga indeks saham, khususnya Indeks Harga Saham Gabungan (IHSG) dan LQ45, menggunakan Support Vector Machine (SVM) dengan variasi moving average: Simple Moving Average (SMA), Exponential Moving Average (EMA), Weighted Moving Average (WMA), dan teknik shifting. Indeks saham mencerminkan pergerakan harga yang dipengaruhi oleh faktor ekonomi, kebijakan, dan kondisi global. Hasil menunjukkan bahwa pada prediksi IHSG, model SVM-EMA memiliki akurasi tertinggi sebesar 98.08% dan MAPE terendah sebesar 2.85%, menunjukkan kemampuan dalam menangkap perubahan harga dengan baik. Pada prediksi LQ45, model SVM WMA mencatat akurasi tertinggi sebesar 96.01%, sedangkan teknik shifting menunjukkan akurasi terendah untuk IHSG (75.13%) dan LQ45 (85.87%), dengan MAPE masing-masing sebesar 3.01% dan 3.87%. Hasil ini menegaskan keunggulan moving average dalam memprediksi harga saham dan pemodelan.

This research evaluates the prediction of stock index prices, specifically the Jakarta Composite Index (IHSG) and LQ45, using Support Vector Machine (SVM) with variations of moving averages: Simple Moving Average (SMA), Exponential Moving Average (EMA), Weighted Moving Average (WMA), and the shifting technique. Stock indices reflect price movements influenced by economic factors, government policies, and global conditions. The results show that for IHSG predictions, the SVM-EMA model achieved the highest accuracy of 98.08% and the lowest MAPE of 2.85%, demonstrating its ability to capture recent price changes effectively. In LQ45 predictions, the SVM-WMA model recorded the highest accuracy of 96.01%, while the shifting technique showed the lowest accuracy for both IHSG (75.13%) and LQ45 (85.87%), with MAPE values of 3.01% and 3.87%, respectively. These findings highlight the superiority of the moving average approach in stock price prediction and modeling.