January 31st, 2019, 12:10 pm
There are plenty of models to forecast volatility from an econometric point of view, look at the garch literature for univariate and multivariate. They work fairly well for daily data, half life is a parameter to be fitted obviously, if picked there needs to be some reasoning.
A pca is not really a forecasting model, but you can use it to create a factor model with it? Again plenty of literature googling on factor models for risk management.
I would start with time varying vol and gaussian distribution.