ARIMA-Based Forecasting of the LQ45 Stock Index: Evidence from 2000–2019

Abstract

Financial markets play a significant role in economic development and investment decision making, particularly in emerging economies where stock indices serve as key indicators of market performance. Among these indices, the LQ45 Stock Index represents one of the most important benchmarks in the Indonesian capital market. However, stock index movements are characterized by high volatility and non-stationary behavior, which complicates the development of reliable forecasting models. This study is motivated by the need for interpretable and robust forecasting approaches that can effectively model long-term market dynamics using historical data. Accordingly, this research proposes an ARIMA-based time series forecasting framework to predict the LQ45 Stock Index using data spanning from 2000 to 2019. The main contribution of this study lies in the comprehensive empirical evaluation of a classical ARIMA model applied to a long-span dataset in an emerging market context, providing insights into its strengths and limitations for stock index forecasting. The proposed model is developed through systematic data preprocessing, stationarity testing, model identification, and diagnostic validation. Forecasting performance is evaluated using standard accuracy metrics, including Mean Absolute Error, Root Mean Squared Error, and Mean Absolute Percentage Error, based on an out-of-sample testing scheme. The results indicate that the ARIMA model is able to capture long-term trends of the LQ45 index and serves as a reliable baseline model, although its performance is limited during periods of high market volatility. Future work may focus on integrating hybrid or machine learning-based approaches and incorporating exogenous variables to improve forecasting accuracy and robustness.