Prediksi Harga Saham Dengan Teknikal Indikator, Faktor Makro Ekonomi, Dan Sentimen Berita Menggunakan Deep Learning

Tyas, Salsabila Mazya Permataning (2024) Prediksi Harga Saham Dengan Teknikal Indikator, Faktor Makro Ekonomi, Dan Sentimen Berita Menggunakan Deep Learning. Masters thesis, Institut Teknologi Sepuluh Nopember.

Text 6025221010-Master_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 April 2026. Download (2MB) | Request a copy

Abstract

Keberhasilan dalam memprediksi harga saham di masa depan dapat menghasilkan keuntungan yang signifikan bagi para investor saham. Penelitian yang sudah dilakukan hanya menggunakan faktor makro ekonomi atau teknikal indikator atau sentimen analisis. Oleh karena itu, penelitian ini mengusulkan prediksi harga saham dengan menggabungkan beberapa indikator yaitu faktor makro ekonomi, teknikal indikator, dan juga sentimen berita. Penelitian ini terdiri dari beberapa tahapan. Pertama, prediksi menggunakan faktor makro ekonomi yaitu inflasi, suku bunga, neraca perdagangan, kurs dollar, ekspor, impor, dan jumlah uang yang beredar. Kedua, prediksi menggunakan teknikal indikator dari historis harga saham. Ketiga, prediksi menggunakan sentimen judul berita. Dalam penelitian ini, dataset judul berita yang digunakan akan dikategorikan berdasarkan masing-masing emitennya untuk memastikan relevansi antara judul berita tersebut dan saham yang diprediksi. Keempat, prediksi menggunakan gabungan seluruh indikator yaitu faktor makro ekonomi, teknikal indikator, dan sentimen judul berita. Dan penelitian ini menggunakan beberapa metode Deep Learning yaitu Gated Reccurent Unit (GRU), Long Short Term Memory (LSTM), dan Bidirectional LSTM (Bi-LSTM) dalam prediksi saham. Penelitian ini dilakukan dengan menggunakan berbagai skenario. Gabungan teknikal indikator, faktor makro ekonomi dan sentimen judul berita dengan metode LSTM dapat meningkatkan performa peramalan dengan nilai R2 0,86, MAPE 2,4, dan RMSE sebesar 132,988.
=================================================================================================================================
Success in predicting future stock prices can produce significant profits for stock investors. Research that has been carried out only uses macroeconomic factors or technical indicators or sentiment analysis. Therefore, this research recommends stock price predictions by combining several indicators, namely macroeconomic factors, technical indicators, and also news sentiment. This study will propose several scenarios for stock prediction. First, the prediction uses macroeconomic factors, namely inflation, interest rates, trade balance, dollar exchange rate, exports, imports, and the amount of money in circulation. Second, predictions use technical indicators from historical stock prices and also indicators of expansion results from stock prices. Third, predictions use news headline sentiment. In this study, the news title dataset used has been categorized based on each issuer to ensure the relevance between the news title and the predicted stock. This is a novelty and aims to improve the accuracy of prediction results. Fourth, the prediction uses a combination of all indicators, namely macroeconomic factors, technical indicators, and news headline sentiment. And this research uses the Deep Learning method, namely Gated Recurrent Unit (GRU), Long Short Term Memory (LSTM), dan Bidirectional LSTM (Bi-LSTM) in stock prediction. This research was conducted using various scenarios. The combination of technical indicators, macroeconomic factors and news headline sentiment using the LSTM method can improve forecasting performance with an R2 value of 0.86, MAPE 2.4, and RMSE of 132.988.

Item Type:	Thesis (Masters)
Uncontrolled Keywords:	Prediksi Saham, Faktor Makro Ekonomi, Teknikal Indikator, Sentimen Berita.Stock predictions, Macroeconomic factors, Technical indicators, News sentiment.
Subjects:	T Technology > T Technology (General) > T174 Technological forecasting
Divisions:	Faculty of Intelligent Electrical and Informatics Technology (ELECTICS) > Informatics Engineering > 55101-(S2) Master Thesis
Depositing User:	Salsabila Mazya Permataning Tyas
Date Deposited:	02 Feb 2024 03:21
Last Modified:	02 Feb 2024 03:21
URI:	http://repository.its.ac.id/id/eprint/105949

Actions (login required)

View Item