quantlab.model.split.split_ranges#

quantlab.model.split.split_ranges(timestamps, windows)[source]#

Cut a backtest’s bars into in-sample and out-of-sample runs.

Parameters:

timestampsnp.ndarray: The sorted bars of the backtest.
windowssequence of (first, last) or None: In-sample windows such as in_sample_window returns, both ends inclusive and compared as exact timestamps; None entries are skipped.

Returns:

in_samplelist of (first, last): The contiguous runs of bars inside any window, in time order.
out_of_samplelist of (first, last): The contiguous runs of the remaining bars, in time order.

Examples

>>> days = pd.bdate_range("2024-01-01", periods=5).values
>>> inside, outside = split_ranges(days, [(days[1], days[2])])
>>> [(str(a)[:10], str(b)[:10]) for a, b in inside + outside]
[('2024-01-02', '2024-01-03'), ('2024-01-01', '2024-01-01'), ('2024-01-04', '2024-01-05')]