numeraire.ClarkWestEvaluator

class numeraire.ClarkWestEvaluator(nw_lags: int = 0)

Bases: object

Clark-West (2007) MSPE-adjusted test of the forecast against its nested benchmark.

The right significance test to pair with OutOfSampleR2Evaluator — plain Diebold-Mariano is undersized against a nested benchmark (the historical mean). Multi-asset outputs aggregate the per-origin adjusted loss difference across assets (the pooled companion of SquaredErrorDiffEvaluator); one asset is the textbook statistic. Emits two rows: cw_t and cw_p (one-sided). Use nw_lags = horizon - 1 for multi-step forecasts.

__init__(nw_lags: int = 0) → None

Methods

__init__([nw_lags])
evaluate(oos_output)

Attributes

requires