Stock Price Prediction Model Based on LSTM Reflecting Interest Rate Fluctuations

Due to the recent impact of inflation, central banks have maintained a policy of raising interest rates, highlighting the growing importance of financial market forecasting that reflects these economic conditions. However, there has been a lack of recent studies in Korea that consider interest rate in stock price prediction. In this study, we aimed to improve prediction accuracy by incorporating interest rates as a variable into stock price prediction models. For this purpose, we utilized the LSTM algorithm and trained the model on historical data from periods of high stock market volatility to predict the KOSPI index. Additionally, we conducted a multi-faceted analysis of the impact of interest rates through market conditions and sector-specific analyses. By employing various evaluation metrics, we performed both stock price prediction and directional movement prediction simultaneously. As a result, the model that incorporated interest rates demonstrated a 48.31% reduction in RMSE, compared to the baseline model, with the performance difference being statistically significant. This study confirms that interest rates are a critical variable in stock price prediction and that models reflecting interest rate fluctuations can enhance predictive performance.