The article improves stock index volatility forecasting by considering different responses to good and bad news and the long-term patterns in volatility. The new model outperforms existing ones in predicting volatility accurately. Including realized power variation as a factor enhances forecasting accuracy, and the model is not easily affected by market noise.