Abstract: The realized volatility is not effective in predicting the future return volatility of the stock market, and the effective predictable information contained in it is not comprehensive. Therefore, in order to improve the predictability of stock return volatility, this paper introduces two important predictors（implicit volatility VIX and WTI （BRT） oil price volatility）, and establishes a "kitchen sink" predictive autoregressive model and "convex combined" predictive autoregressive model. The empirical results show that the "kitchen sink" predictive autoregressive model that combines the implied volatility VIX and WTI（BRT） oil price volatility has a good predictive effect. To further prove the reliability of this conclusion, the paper also adds economic cycles or changes conditions such as macroeconomic and financial variables. The results show that, compared to the WTI（BRT） oil price volatility, the implied volatility VIX is more effective for the predictability of other macroeconomic and financial variables. Finally, based on the different lag lengths and the dimensions of different macroeconomic variables, this article conducts robustness tests on various predictive regression models. The test results show that the autoregressive model combined with implied volatility VIX and WTI（BRT） oil price volatility is more effective and stable in predicting stock return volatility.

Key words: implied volatility（VIX）, WTI（BRT） oil price volatility, autoregressive model（AR）, robustness