Description of concepts, models and methods

Daily out-of-sample risk forecasts for a range of assets. Estimation window is 1000 days, and probability is 99%, with portfolio value 1000 in the native currency. Returns are log returns, $y_t=\log(p_t/p_{t-1})$.

The two risk measures that are estimated are Value-at-Risk (VaR) and expected shortfall (ES), both at the 99% probability.

These measures are described here.

The statistical models used in the forecasting can be seen here.

The models are estimated with code from Financial Risk Forecasting.