Realized volatility models and augmented GARCH models with certain estimation methods are better at predicting Value at Risk (VaR) compared to other models. These models provide more accurate VaR forecasts and help in allocating regulatory capital more efficiently, especially during times of high market volatility.