BacktestResult.errors

property BacktestResult.errors: NDArray[float64]

(T, H, k) forecast errors point - realized.

The sign convention is forecast minus outcome, so a positive error is an over-forecast. Recomputed on each access from the two stored arrays.

Example

>>> import numpy as np
>>> from cultivars.multivariate.reduced_form.vector_autoregression import VAR
>>> y = np.random.default_rng(0).standard_normal((130, 2))
>>> record = Backtest(y, lambda w: VAR(w, order=1).fit(), horizons=3, start=100).run()
>>> record.errors.shape
(28, 3, 2)
>>> bool(np.allclose(record.errors, record.point - record.realized))
True