Backtest.origins#

property Backtest.origins: NDArray[int64][source]#

Observations available at each origin the schedule will evaluate.

The arithmetic sequence start, start + step, ... up to the last value that still leaves horizons rows to forecast, \(n - H\); its length is the \(T\) of the record run() returns.

Example

>>> import numpy as np
>>> y = np.random.default_rng(0).standard_normal((130, 2))
>>> Backtest(y, lambda w: w, horizons=2, start=100, step=7).origins
array([100, 107, 114, 121, 128])