Abstract:

This study aims to analyze the accuracy of the SVM model to predict stocks price. Herein, a combination of forecast period and input period technical indicators is used with longer time frames with a support vector machine application to forecast future stock price movements in the Indonesian market. A combination of 28 forecasting periods and 30 input period technical indicators was used. Stock transaction data from 31 companies listed on the Indonesian stock exchange and actively traded between March, 2006 and February, 2018 were used. Results show that the highest system performance does not occur when the input period technical indicator is approximately equal to the forecast period. Although, the performance results differ among company stocks. However, the SVM prediction model provides greater profit than buy and hold strategy.