Prediksi Harga Saham PT Bank Syariah Indonesia Tbk Berdasarkan Sentimen Pasar Menggunakan IndoBERT dan Long Short-Term Memory

Fathony, Nyimas Syifa Azahraa Dzulia Putri (2024) Prediksi Harga Saham PT Bank Syariah Indonesia Tbk Berdasarkan Sentimen Pasar Menggunakan IndoBERT dan Long Short-Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 2043201025-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2026. Download (2MB) | Request a copy

Abstract

Lembaga keuangan syariah di Indonesia berpotensi untuk berkembang pesat mengingat banyaknya penduduk muslim di Indonesia. Potensi perkembangan tersebut mewujudkan penggabungan tiga bank syariah BUMN menjadi PT Bank Syariah Indonesia Tbk (BSI) dengan kode saham BRIS. Berita keberhasilan penggabungan ini dinantikan oleh para investor dan memberikan dampak signifikan terhadap harga saham BSI. Di pasar modal, sentimen pasar memengaruhi fluktuasi harga saham sehingga penelitian ini meramalkan harga saham BSI dengan mempertimbangkan sentimen pasar. Analisis sentimen dilakukan dengan IndoBERT, kemudian informasi sentimen serta data histori harga saham digunakan untuk peramalan menggunakan Long Short-Term Memory (LSTM). Terdapat tiga skenario yang dibandingkan: harga saham BRIS tanpa mempertimbangkan sentimen pasar, harga saham dipengaruhi oleh sentimen hari ini, dan harga saham dipengaruhi oleh sentimen hari ini serta hari sebelumnya. Hasil analisis menunjukkan bahwa mayoritas berita BRIS di portal CNBC bersentimen neutral, yaitu 64.9% untuk berita BRIS periode ke-t dan 64.8% untuk berita periode t-1. Model IndoBERT menunjukkan akurasi prediksi sentimen neutral dan positive yang tinggi, tetapi kurang baik dalam memprediksi sentimen negative. Pada pemodelan prediksi harga saham BRIS, ditemukan bahwa harga saham BRIS mengalami fluktuasi signifikan dengan tren penurunan akibat penilaian pasar yang menganggap saham overvalued dan tren kenaikan diduga karena kinerja bank yang baik. Model terbaik adalah model prediksi harga saham BRIS tanpa mempertimbangkan sentimen pasar. Model menunjukkan hasil prediksi yang baik dengan MAPE 3.564% dan RMSE 108.095, serta meremalkan tren penurunan harga saham yang konsisten untuk 7 hari ke depan.
=================================================================================================================================
Islamic financial institutions in Indonesia have the potential for rapid growth given the large Muslim population in the country. This growth potential led to the merger of three state-owned Islamic Banks into PT Bank Syariah Indonesia Tbk (BSI), with the stock code BRIS. The news of this successful merger was eagerly awaited by investors and had a significant impact on BSI's stock price. In the capital market, market sentiment influences stock price fluctuations. Therefore, this study forecasts BSI's stock price by considering market sentiment. Sentiment analysis was conducted using IndoBERT, and then sentiment information and historical stock price data were used for forecasting using Long Short-Term Memory (LSTM). Three scenarios were compared: BRIS stock prices without considering market sentiment, stock prices influenced by today's sentiment, and stock prices influenced by today's and the previous day's sentiment. The analysis results showed that the majority of BRIS news on the CNBC portal was neutral sentiment, with 64.9% for BRIS news in the t period and 64.8% for news in the t-1 period. The IndoBERT model showed high accuracy in predicting neutral and positive sentiments but was less effective in predicting negative sentiment. In the BRIS stock price prediction modeling, it was found that BRIS stock prices experienced significant fluctuations, with a downward trend due to market assessments that the stock was overvalued and an upward trend likely due to good bank performance. The best model was the BRIS stock price prediction model without considering market sentiment. The model showed good prediction results with MAPE of 3.564% and RMSE of 108.095, and forecasted a consistent downward trend in stock prices for the next 7 days.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Bank Syariah Indonesia, IndoBERT, LSTM
Subjects:	H Social Sciences > HA Statistics > HA30.3 Time-series analysis
Divisions:	Faculty of Vocational > 49501-Business Statistics
Depositing User:	Nyimas Syifa Azahraa Dzulia Putri Fathony
Date Deposited:	20 Aug 2024 08:05
Last Modified:	20 Aug 2024 08:05
URI:	http://repository.its.ac.id/id/eprint/114761

Actions (login required)

View Item