Perbandingan Prediksi Pergerakan Data IHSG Menggunakan Feed Forward Neural Network dan Long Short Term Memory

Pamungkas, Taufiq Hario (2023) Perbandingan Prediksi Pergerakan Data IHSG Menggunakan Feed Forward Neural Network dan Long Short Term Memory. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 06311940000048-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only until 1 October 2025. Download (3MB) | Request a copy

Abstract

Investasi saham memiliki potensi keuntungan yang tinggi dibandingkan dengan instrumen investasi lainnya. Namun, hal ini juga berarti terdapat risiko kerugian yang sebanding. Dalam investasi saham, terdapat dua risiko utama yang perlu diperhatikan. Pertama, risiko capital loss, di mana investor membeli saham dengan harga tertentu dan menjualnya di bawah harga beli karena berbagai faktor yang mempengaruhi pasar. Risiko kedua adalah risiko kebangkrutan perusahaan yang menerbitkan saham, yang dapat mengakibatkan investor kehilangan potensi keuntungan. Untuk menghindari risiko-risiko tersebut, penting bagi investor saham untuk memahami fluktuasi harga saham yang tercermin dalam Indeks Harga Saham Gabungan (IHSG). IHSG menjadi acuan dalam memprediksi pergerakan harga saham, sehingga investor dapat menghindari kerugian dalam pasar modal. IHSG adalah data time series yang dapat diperkirakan menggunakan metode Artificial Neural Network (ANN), yaitu teknik pengolahan informasi yang terinspirasi oleh cara kerja sistem saraf biologis, khususnya dalam pemrosesan informasi oleh sel otak manusia. Dalam penelitian ini, dilakukan perbandingan antara dua model ANN yang sering digunakan dalam pasar modal, yaitu Feed Forward Neural Network (FFNN) dan Long Short Term Memory (LSTM). FFNN merupakan model ANN yang paling sederhana, sedangkan LSTM merupakan pengembangan dari model Recurrent Neural Network (RNN) yang memungkinkan neuron untuk mengingat informasi dalam jangka waktu yang lebih panjang. Hasil penelitian menunjukkan bahwa model FFNN (3-6-1) memiliki performa yang lebih baik dalam memprediksi pergerakan IHSG, dengan nilai MAPE sebesar 0,652 % , RMSE sebesar 60,951, dan MAE sebesar 45,245. Dengan prediksi IHSG berdasarkan model FFNN dapat membantu investor dalam pengambilan keputusan investasi yang lebih baik dan manajemen risiko yang lebih efektif di pasar saham karena dari hasil penelitian menunjukan FFNN lebih baik dari LSTM.
================================================================================================================================
Stock investment has a high profit potential compared to other investment instruments. However, this also means that there is a comparable risk of loss. In stock investing, there are two main risks that need attention. First, the risk of capital loss, where investors buy shares at a certain price and sell them below the purchase price due to various factors that affect the market. The second risk is the risk of bankruptcy of the company issuing the shares, which can result in investors losing potential profits. To avoid these risks, it is important for stock investors to understand stock price fluctuations as reflected in the Jakarta Composite Index (IHSG). JCI is a reference in predicting stock price movements, so that investors can avoid losses in the capital market. IHSG is time series data that can be estimated using the Artificial Neural Network (ANN) method, which is an information processing technique inspired by the workings of the biological nervous system, particularly in information processing by human brain cells. In this study, a comparison was made between the two ANN models that are often used in the capital market, namely Feed Forward Neural Network (FFNN) and Long Short Term Memory (LSTM). FFNN is the simplest ANN model, while LSTM is a development of the Recurrent Neural Network (RNN) model which allows neurons to remember information in a longer period of time. The results showed that the FFNN model (3-6-1) had better performance in predicting the IHSG movement, with a MAPE value of 0.652 %, RMSE of 60.951, and MAE of 45.245. With IHSG predictions based on the FFNN model it can assist investors in making better investment decisions and more effective risk management in the stock market because the research results show that FFNN is better than LSTM.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	Feed Forward Neural Network (FFNN), Composite Stock Price Index, Long Short Term Memory (LSTM), Indek Harga Saham Gabungan,
Subjects:	Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science)
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Actuaria > 94203-(S1) Undergraduate Thesis
Depositing User:	Taufiq Hario Pamungkas
Date Deposited:	01 Aug 2023 02:26
Last Modified:	01 Aug 2023 02:26
URI:	http://repository.its.ac.id/id/eprint/100850

Actions (login required)

View Item