split_sessions¶

Defined in fynance.data

split_sessions(X, ts, open='09:30', close='16:00', utc_offset=0.0, weekdays_only=True)[source]

Split X into one chunk per trading session.

Parameters:

Xndarray: Array aligned with ts along its first axis.
tsndarray of int64 or float64: Non-decreasing epoch-second (UTC) timestamps, shape (n,).
open, close, utc_offset, weekdays_only: See session_mask.

Returns:

list of ndarray: One slice of X per session, in session order. Bars outside every session (session_id == -1) are skipped, so concatenating the result along axis 0 reproduces X[session_mask(ts, ...)].

Raises:

ValueError: Same as session_mask.

Examples

>>> import numpy as np
>>> from fynance.data.sessions import split_sessions
>>> ts = np.array([9 * 3600, 10 * 3600, 17 * 3600, 33 * 3600, 34 * 3600])
>>> X = np.arange(5)
>>> split_sessions(X, ts, open="09:30", close="16:00")
[array([1]), array([4])]