ABSTRAKSI: Penelitian ini bertujuan untuk memprediksi trend kenaikan Indeks Harga Saham Gabungan (IHSG) menggunakan metode Support Vector Machine (SVM). Untuk memprediksi kenaikan IHSG menggunakan empat fitur input yaitu Indeks Dowjones Industrial Average (DJIA), iShares MSCI Indonesia Investable Market Index Fund (EIDO), Nilai Tukar Rupiah terhadap Dolar Amerika Serikat (Kurs), dan IHSG pada periode Mei 2010 sampai dengan Juni 2014. Penggunaan Kernel trick pada metode SVM diharapkan dapat diklasifikasikan secara linear dengan fungsi KernelGaussian Radial Basis Function (RBF) dan Polynomial pada non-linear separable data, dimana pengolahan data fitur input menggunakan 5 K-Fold Cross-Validation dengan tujuan mencari validasi terbaik. Hasil penelitian menunjukkan bahwa metode SVM dengan menggunakan fungsi KernelRBF belum optimal dalam memprediksi trend kenaikan IHSG, hal ini dibuktikan dengan rata-rata akurasi testing terbaik diperoleh 62.8963 %pada kombinasi featureinput IHSG terhadap Indeks EIDO (iShares MSCI Indonesia Investable Market Index Fund).KATA KUNCI: IHSG, Support Vector Machine,Feature Space, Hyperplane, Kernel Trick.ABSTRACT: This study aims to predict the rising trend of IHSG (formerly known as Jakarta Composite Index, JCI or JSX Composite) using Support Vector Machine method.To predict the rising trend of IHSG used four feature input, there are DowJones Industrial Average Index (DJIA), iShares MSCI Indonesia investable Market Index Fund (EIDO), Exchange Rate against the United States Dollar (Exchange), and JCI in the period May 2010 until June 2014. The use of kernel trick on SVM method is expected to be classified linearly with RBF kernel and polynomial functions on non-linear separable data, which the data processing of feature inputusing 5 K-Fold Cross-Validation in order to find the best validation. The results showed that the SVM method using Gaussian kernel function Radial Basis Function (RBF) is not optimal in predicting the trend of rising in JCI, this is evidenced by the average of the best testing accuracy of 62.8963 % on a combination offeature input IHSG against EIDO EIDO (iShares MSCI Indonesia Investable Market Index Fund).KEYWORD: IHSG, Support Vector Machine, Feature Space, Hyperplane, Kernel Trick.