Meiviananda, Muslikh Annur (2020) Peramalan Harga Saham Berdasarkan Faktor Makroekonomi Menggunakan Arsitektur Long Short-Term Memory (Studi Kasus: PT Astra International Tbk.). Other thesis, Institut Teknologi Sepuluh Nopember.
Preview |
Text
05211640000092-Undergraduate_Thesis.pdf Download (3MB) | Preview |
Abstract
Investasi saham merupakan hal yang cukup menjanjikan untuk saat ini dan di masa yang akan datang. Besarnya benefit yang ditawarkan mendorong berbagai kalangan untuk mempertimbangkan investasi pada saham perusahaan. Hal tersebut dibuktikan dengan adanya peningkatan jumlah investor di Bursa Efek Indonesia (BEI) sebanyak 26% dari Juli 2015 hingga Juli 2016. Meskipun dirasa menjanjikan, terdapat banyak indikator yang mempengaruhi harga saham seperti sentimen perusahaan dan faktor makroekonomi yang beragam. Indikator seperti nilai tukar USD/IDR termasuk dalam faktor makroekonomi yang mempengaruhi harga saham perusahaan di beberapa kasus. Salah satu perusahaan dengan sentimen positif di mata masyarakat adalah PT Astra International Tbk (Astra). Astra merupakan perusahaan multinasional diversivikasi dengan tujuh segmen usaha. Astra sudah terdaftar di BEI sejak tahun 1990. Kategori saham Astra termasuk blue-chip stocks yang mana perusahaan dengan label tersebut adalah perusahaan bereputasi tinggi dengan pendapatan stabil dan konsisten dalam membayar dividen. Peramalan harga saham Astra akan membantu para investor yang ingin menanamkan modal dalam ketepatan pengambilan keputusan. Pengambilan keputusan yang baik membuat para investor dapat maksimalkan benefit dari investasi saham yakni capital gain dan dividen.
Sudah banyak metode untuk meramalkan data saham yang bersifat non-linear salah satunya Recurrent Neural Network (RNN). RNN merupakan jenis arsitektur Neural Network yang pemrosesannya dipanggil berulang-ulang. RNN terbukti sangat baik dalam memproses input berupa data time series seperti data saham. Pada penelitian kali ini, diajukan salah satu solusi yang merupakan arsitektur RNN yakni Long Short-Term Memory (LSTM). LSTM merupakan pembelajaran berbasis gradien dengan tiga gates yakni input, forget dan output.
Performa model LSTM pada peramalan harga saham Astra menggunakan faktor makroekonomi berupa nilai tukar USD/IDR menunjukkan performa yang sangat baik dibuktikan dengan Mean Absolute Percentage Error (MAPE) yang kecil. Model dengan akurasi terbaik yang dihasilkan menunjukkan peramalan saham tanpa faktor makroekonomi menghasilkan MAPE 1,3465% dan dengan faktor makroekonomi menghasilkan MAPE 1,3512%. Hasil peramalan juga menunjukkan faktor makroekonomi tidak berdampak pada performa yang lebih baik.
================================================
Stock investment is quite promising for now and in the future. The amount of benefits offered encourages various groups to consider investing in company shares. This is evidenced by an increase in the number of investors in the Indonesia Stock Exchange (IDX) by 26% from July 2015 to July 2016. Although it is considered promising, there are many indicators that affect stock prices such as corporate sentiment and various macroeconomic factors. Indicators such as the USD / IDR exchange rate are included in macroeconomic factors that affect the company's stock price in some cases. One company with positive sentiment in the eyes of the public is PT Astra International Tbk (Astra). Astra is a diversified multinational company with seven business segments. Astra has been listed on the IDX since 1990. Astra's share category includes blue-chip stocks in which companies with that label are highly-reputed companies with stable income and are consistent in paying dividends. Astra's stock price forecasting will help investors who want to invest capital in the accuracy of decision making. Good decision making allows investors to maximize the benefits of stock investments, namely capital gains and dividends.
There have been many methods for predicting non-linear stock data, one of which is the Recurrent Neural Network (RNN). RNN is a type of Neural Network architecture whose processing is called repeatedly. RNN has proven to be very good in processing input in the form of time series data such as stock data. In this study, a solution which is an RNN architecture is proposed, namely Long Short-Term Memory (LSTM). LSTM is a gradient-based learning with three gates namely input, forget and output.
The LSTM model's performance on Astra's stock price forecasting using macroeconomic factors in the form of the USD/IDR exchange rate shows excellent performance as evidenced by the small Mean Absolute Percentage Error (MAPE). The best model produced shows stock forecasting without macroeconomic factors produces MAPE 1.3465% and with macroeconomic factors produces MAPE 1.3512%. Forecasting results also show that macroeconomic factors do not have an impact on better performance.
| Item Type: | Thesis (Other) |
|---|---|
| Uncontrolled Keywords: | Peramalan, Harga Saham, Makroekonomi, Long Short-Term Memory, Mean Absolute Percentage Error, Forecasting, Stock Price, Macroeconomic, Long Short-Term Memory, Mean Absolute Percentage Error. |
| Subjects: | H Social Sciences > HB Economic Theory > Economic forecasting--Mathematical models. T Technology > T Technology (General) > T174 Technological forecasting |
| Divisions: | Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Information System > 57201-(S1) Undergraduate Thesis |
| Depositing User: | Muslikh Annur Meiviananda |
| Date Deposited: | 01 Sep 2020 03:05 |
| Last Modified: | 23 May 2023 05:01 |
| URI: | http://repository.its.ac.id/id/eprint/78644 |
Actions (login required)
 |
View Item |