Analisis Kinerja Support Vector Machine dan Moving Averages Convergence Divergence Untuk Saham-Saham Perbankan Indonesia

This study discusses the Machine Learning algorithm with technical indicator features in predicting the movement of Indonesian banking sector stocks. Many people seek profit in Indonesian banking stocks because most of them have good fundamentals but have high volatility. The strategy that can be used is the Support Vector Machine (SVM) algorithm with the Moving Averages Convergence Divergence (MACD) technical indicator feature. The SVM algorithm is used because it can process stock price movement data and technical indicators which tend to be complex. This research was conducted with the aim of contributing to the development of a machine learning-based stock prediction model that can be used by academics and financial practitioners. The research stages are collecting historical data on Infobank15 stock movements, data cleaning, training and testing the SVM model, then backtesting. The research methodology includes data processing using Python, training and testing the SVM model, then trading simulation with an initial capital of IDR 100 million. The kernels tested include RBF, Polynomial, and Sigmoid. Model performance is evaluated based on return, win rate, profit ratio, Sharpe ratio, maximum drawdown, risk-reward ratio, and accuracy rate. Historical data of Infobank15 stock is used in this study where the dataset is historical data from 2019-2024 for training and testing the model and historical data from 2024 for backtesting. Based on the experimental results obtained, it can be concluded that the combination of the SVM model and the MACD indicator yields favourable outcomes. The kernel that provides the best overall results is the Polynomial kernel. Penelitian ini membahas penerapan Machine Learning dengan fitur indikator teknikal dalam memprediksi pergerakan saham sektor perbankan Indonesia. Banyak masyarakat mencari keuntungan di saham perbankan Indonesia karena sebagian besar memiliki fundamental yang baik, namun memiliki volatilitas yang tinggi. Strategi yang dapat digunakan yaitu algoritma Support Vector Machine (SVM) dengan fitur indikator teknikal Moving Averages Convergence Divergence (MACD). Algoritma SVM dipakai karena dapat mengolah data-data pergerakan harga saham dan indikator teknikal yang di mana cenderung kompleks. Riset ini dilakukan dengan tujuan berkontribusi pada pengembangan model prediksi saham berbasis Machine Learning yang dapat digunakan oleh akademisi dan praktisi keuangan. Tahapan risetnya yaitu pengumpulan data historis pergerakan saham Infobank15, pembersihan data, pelatihan dan pengujian model SVM, kemudian backtesting. Metodologi risetnya meliputi pengolahan data menggunakan Python, pelatihan dan pengujian model SVM, kemudian simulasi trading. Kernel yang diuji antara lain Radial Basis Function, Polynomial, dan Sigmoid. Kinerja model dievaluasi berdasarkan return, win-rate, profit ratio, sharpe ratio, maximum drawdown, risk-reward ratio, dan accuracy rate. Data historis saham Infobank15 digunakan dalam penelitian ini adalah tahun 2019-2024 untuk pelatihan dan pengujian model serta data historis 2024 untuk backtesting. Dari hasil percobaan yang telah dilakukan, dapat disimpulkan bahwa kombinasi model SVM dan indikator MACD memberikan hasil yang baik. Kernel yang memberikan hasil terbaik secara keseluruhan adalah Polynomial.