ANALISIS TREN HARGA PENUTUPAN SAHAM BEBERAPA BANK MENGGUNAKAN MODEL DERET WAKTU UNIVARIAT DAN MULTIVARIAT

• Main Author: Stewart Leonardo, Justin
• Format: Theses
• Language: Indonesia
• Online Access: https://digilib.itb.ac.id/gdl/view/83920
• Institution: Institut Teknologi Bandung

This study aims to analyze the trend of stock closing prices of three major banks in Indonesia, namely BCA, Mandiri, and BRI, during the period from January to July 2024. The analysis was carried out using univariate time series models such as AR, MA, ARMA, and ARIMA, as well as the multivariate Vector Autoregressive (VAR) model. Daily stock closing price data was obtained from Yahoo Finance, which was then processed through a series of preprocessing steps including stationarity tests using Augmented Dickey-Fuller (ADF) and differentiation to achieve stationarity. After the data was divided into training data and testing data, a prediction model was built and tested to evaluate the prediction accuracy. The results showed that the ARIMA model tends to be superior when compared to the VAR model in terms of prediction accuracy based on RMSE and MAPE metrics, especially for individual stock data. However, the VAR model is more effective in capturing dynamic interactions between stocks, although with slightly lower prediction accuracy than the univariate model. The novelty of this study is a comprehensive comparison between the performance of univariate and multivariate models in the context of banking stock prediction in Indonesia. This research makes an important contribution to the economics and finance literature by providing guidance for investors and analysts in selecting the prediction model that best suits their needs