Data science approach to stock prices forecasting in Indonesia during Covid-19 using Long Short-Term Memory (LSTM)

Area of Science:

• Data Science
• Financial Forecasting
• Machine Learning

Background:

• Stock market forecasting is crucial for financial decision-making due to inherent uncertainties.
• The Indonesian stock market experienced significant volatility, particularly during the COVID-19 pandemic.
• Accurate stock price prediction is vital for investors and financial institutions.

Purpose of the Study:

• To propose a data science model for forecasting stock prices on the Indonesian exchange.
• To leverage statistical computing with R and Long Short-Term Memory (LSTM) for enhanced prediction accuracy.
• To analyze the impact of the COVID-19 pandemic on stock prices and evaluate forecasting model performance.

Main Methods:

• Utilized big data from Indonesian banks (BCA, Bank of Mandiri) sourced from Yahoo Finance.
• Applied data science techniques for data visualization and analysis.
• Implemented a Long Short-Term Memory (LSTM) network for stock price prediction, focusing on Open, High, Low, and Closing (OHLC) values.

Main Results:

• The Indonesian composite stock price index (IHSG) plunged 28% in early 2020 due to the COVID-19 pandemic.
• Share prices of major sectors like banking and consumer goods reached record lows in March 2020.
• The proposed LSTM model demonstrated effective visualization and prediction of OHLC prices.

Conclusions:

• Data science provides valuable tools for financial data visualization.
• The proposed LSTM model achieved a high accuracy of 94.57% for short-term stock price forecasting (1-year data).
• Training with a higher epoch count on shorter-term data yielded better predictive performance compared to longer-term data.