Construct loose bounds for conditional variances.
These bounds are used in parameter estimation to ensure
that the log-likelihood does not produce NaN values.
- Parameters:
- resids: ndarray
Approximate residuals to use to compute the lower and upper bounds
on the conditional variance
- power: float =
2.0
Power used in the model. 2.0, the default corresponds to standard
ARCH models that evolve in squares.
- Returns:
var_bounds – Array containing columns of lower and upper bounds with the same
number of elements as resids
- Return type:
numpy.ndarray