arch.univariate.ARCHInMean.compute_param_cov¶

ARCHInMean.compute_param_cov(params: ndarray[tuple[int], dtype[float64]], backcast: float | ndarray[tuple[int], dtype[float64]] | None = None, robust: bool = True) → ndarray[tuple[int, ...], dtype[float64]]¶

Computes parameter covariances using numerical derivatives.

Parameters:¶

params: ndarray[tuple[int], dtype[float64]]¶

Model parameters

backcast: float | ndarray[tuple[int], dtype[float64]] | None = None¶

Value to use for pre-sample observations

robust: bool = True¶

Flag indicating whether to use robust standard errors (True) or classic MLE (False)