Forecasting the volatility of educational firms based on HAR model and LSTM models considering sentiment and educational policy

This study aims to investigate the impact of sentiment and policy on the volatility of educational stock prices by using HAR (Heterogeneous Auto Regressive) and LSTM (Long Short-Term Memory) models. We construct a weighted educational index volatility composed of nine publicly traded educational companies from the Shenzhen Stock Exchange and Shanghai Stock Exchange, and analyze the impact of sentiment and policy variables on the volatility of educational stock prices. We use OLS regression models and LSTM prediction models to analyze the data by developing various of models to investigate the impact of sentiment, education policies and their intersection effect. The empirical results show that the sentiment index and policy index have significant impacts on different time horizons of educational stock price volatility. The LSTM model confirms the effectiveness of including sentiment and policy variables in predicting educational stock price volatility. These findings carry several practical implications, particularly for investors, education-listed companies, and policymakers. And this study contributes to the literature by providing new evidence on the impact of sentiment and policy on the volatility of educational stock prices and by demonstrating the usefulness of combining HAR and LSTM models in predicting stock price volatility.