Machine-learning-vs.-autoregression-forecasting-ETF-realized-volatilty
A machine learning approach of recurrent neural networks outperforms heterogenous autoregressive (HAR) models in out-of-sample forecasting of realized volatility in customized ETFs.
