Type Alias SimulationRequest

GARCH-X (params with δ): the x that enter the variances, x_T for step 1's and x_{T+k} for step k + 1's — steps values, or one held throughout. Without a state to start from, the paths start at (ω + δ·mean(x))/(1 − P).

Where the innovations come from: absent, the parametric family the params name; { bootstrap: z }, drawn with replacement from z — a fit's standardised residuals (standardizedResiduals), as filtered historical simulation draws them. ν and λ are then not used for the draws.

Where the recursion stands: s and e of the last observed return. With both, step 1's variance is the one-step forecast; without, the paths start at the unconditional variance. A FitAt spreads into the request.

With nu, the innovations are Student-t standardised to unit variance, with lambda too Hansen's skewed t (garch and gjr only).

The same seed gives the same paths, on any platform.