Stocks are one of the popular investment instruments in Indonesia, but they also come with high risks. According to a report from PT Central Securities Depository Indonesia (KSEI) on November 21, 2022, the number of investors in the Indonesian capital market has reached 10 million people, followed by 9.3 million mutual fund investors and 803 thousand investors in government securities. Therefore, a method is needed to accurately predict stock prices. This research aims to apply a Machine Learning method based on Recurrent Neural Network (RNN) to predict stock prices on the Indonesia Stock Exchange (IDX) using financial ratio variables as features. The study uses stock data on the IDX from the Kompas100 stock index from 2008 to 2022. The data is then converted, cleaned, and divided into training and testing data. The model is trained with the training data and tested with the testing data. To measure the model's performance, this research uses the Root Mean Square Error (RMSE) metric. The study finds that the best-performing model produces an RMSE of 14.08. Additionally, the research also discovers that the best combination of financial ratio variables for the machine learning model includes ROA, ROE, PBV, DER, EPS, and PER, compared to combinations such as EPS, ROE, PER, and DER, or ROA, ROE, EPS, and PBV. In conclusion, this study suggests that the Machine Learning method based on RNN can be used to predict stock prices on the IDX using financial ratio variables as features.
Read full abstract