1. Concerning realized correlation, RCM-NBEATSx produces on average 13.38% and 16.43% more accurate forecasts regarding RMSE and MAE respectively. Moreover, both statistical tests confirm the forecast accuracy superiority of RCM-NBEATSx. Apart from DM test (MSE) for DHEAVY and MCS test (MAE) for LogM-HAR, all null hypothesis can be rejected considering the conservative threshold of 0.01. Additionally, given the all statistical and error measures results;23% more accurate forecasts considering RMSE, MAE, and QLIKE respectively
2. The Distribution of Stock Return Volatility
3. Forecasting multivariate realized stock market volatility;G H Bauer;Journal of Econometrics,2011
4. A Capital Asset Pricing Model with Time-Varying Covariances;T Bollerslev;Journal of Political Economy,1988
5. Forecasting large scale conditional volatility and covariance using neural network on GPU;X Cai;The Journal of Supercomputing,2012