Automated Volatility Forecasting
开发了一个自动化系统，利用超过100个特征和五种机器学习算法预测波动率，在标普100股票上表现优于现有模型，并可通过超参数迁移学习扩展到标普500，统计改进转化为显著的年化收益。
We develop an automated system to forecast volatility by leveraging more than 100 features and five machine learning algorithms. Considering the universe of S&P 100 stocks, our system results in superior out-of-sample volatility forecasts compared with existing risk models across forecast horizons. We further demonstrate that our system remains robust to different specifications and is scalable to a broader S&P 500 stock universe via hyperparameter transfer learning. Finally, the statistical improvement in volatility forecasts translates into significant annual returns from a cross-sectional variance risk premium strategy. This paper was accepted by Lin William Cong, finance. Supplemental Material: The online appendix and data files are available at https://doi.org/10.1287/mnsc.2023.01520 .