Analisis Pola Grafik Saham Anggota IDX30 Dengan Metode Convolutional Neural Network

Gusman, Imam Nur Rizky (2025) Analisis Pola Grafik Saham Anggota IDX30 Dengan Metode Convolutional Neural Network. Other thesis, Institut Teknologi Sepuluh Nopember.

Text 5003211121-Undergraduate_Thesis.pdf - Accepted Version Restricted to Repository staff only Download (7MB) | Request a copy

Abstract

Pasar modal Indonesia mengalami perkembangan pesat dengan peningkatan jumlah investor muda yang membutuhkan alat analisis yang efisien. Analisis teknikal konvensional seringkali rentan terhadap kesalahan manusia dalam mengenali pola grafik saham. Penelitian ini bertujuan untuk menerapkan model Convolutional Neural Network (CNN) untuk mengklasifikasikan pola grafik saham, khususnya pola head and shoulders, inverse head and shoulders, double top, dan double bottom, dari data time series di TradingView, yang dilakukan generate menjadi gambar candlestick chart saham anggota IDX30. Dengan menerapkan CNN, diharapkan dapat meningkatkan efisiensi dalam analisis teknikal, sehingga membantu investor dalam pengambilan keputusan yang lebih baik di pasar modal Indonesia. Model CNN untuk melakukan klasifikasi pola candlestick berdasarkan gambar yang diinput. Data time series berupa open, high, low, close (OHLC) digunakan untuk generate gambar pola saham berupa candlestick chart menggunakan matplotlib. Gambar hasil generate digunakan sebagai input model. Model CNN dengan arstitektur MobileNetV2 memiliki akurasi terbaik sebesar 91,67% pada data gambar RGB dan 90,38% pada data gambar grayscale. Model CNN dengan arsitektur MobileNetV2 memiliki kinerja yang lebih baik dari model CNN baseline dalam analisis pola grafik saham, dengan akurasi sebesar 83,33%. Model CNN baseline juga memiliki indikasi overfitting dengan selisih akurasi training dan testing yang tinggi. Hasil model terbaik diimplementasikan ke dalam aplikasi berbasis web (webapps) yang memanfaatkan model CNN untuk klasifikasi pola candlestick berdasarkan gambar yang di upload. Sebagai penunjang kebutuhan analisis, webapps juga memiliki fitur untuk generate gambar pola saham dalam bentuk candlestick chart, serta fitur untuk mengunduh data time series saham di Bursa Efek Indonesia. Penelitian ini diharapkan dapat memberikan kontribusi dalam pengembangan teknologi analisis saham di pasar modal Indonesia, khususnya pada pengembangan metode machine learning.
===================================================================================================================================
The Indonesian capital market has experienced rapid growth, with an increasing number of young investors requiring efficient analytical tools. Conventional technical analysis is often prone to human error in identifying stock chart patterns. This research aims to apply the Convolutional Neural Network (CNN) model to classify stock chart patterns, particularly the head and shoulders, inverse head and shoulders, double top, and double bottom patterns, from time series data on TradingView, which is generated into candlestick chart images of IDX30 stocks. By implementing CNN, it is expected to improve the efficiency of technical analysis, thus helping investors in making better decisions in the Indonesian capital market. The CNN model classifies candlestick patterns based on input images. Time series data in the form of open, high, low, and close (OHLC) values are used to generate stock pattern images in the form of candlestick charts using matplotlib. The generated images are then used as input to the model. The MobileNetV2 CNN model achieved the best accuracy of 91.67% on RGB image data and 90.38% on grayscale image data. The MobileNetV2 CNN model outperformed the baseline CNN model in stock chart pattern analysis, with an accuracy of 83.33%. The baseline CNN model also showed signs of overfitting with a significant gap between training and testing accuracy. The best model was implemented into a web-based application (web app) that utilizes the CNN model for classifying candlestick patterns based on uploaded images. To support analytical needs, the web app also includes features to generate stock pattern images in the form of candlestick charts, as well as the ability to download time series data of stocks listed on the Indonesia Stock Exchange (IDX). This research is expected to contribute to the development of stock analysis technology in the Indonesian capital market, particularly in the advancement of machine learning methods.

Item Type:	Thesis (Other)
Uncontrolled Keywords:	CNN, MobileNetV2, IDX30, Candlestick Chart, Analisis Teknikal CNN, MobileNetV2, IDX30, Candlestick Chart, Technical Analysis
Subjects:	Q Science > Q Science (General) > Q337.5 Pattern recognition systems Q Science > QA Mathematics > QA76.87 Neural networks (Computer Science) T Technology > TA Engineering (General). Civil engineering (General) > TA1637 Image processing--Digital techniques. Image analysis--Data processing.
Divisions:	Faculty of Science and Data Analytics (SCIENTICS) > Statistics > 49201-(S1) Undergraduate Thesis
Depositing User:	Imam Nur Rizky Gusman
Date Deposited:	23 Jul 2025 08:44
Last Modified:	23 Jul 2025 08:44
URI:	http://repository.its.ac.id/id/eprint/120672

Actions (login required)

View Item