Hi CrunchDAO team,
I had a question regarding the labeling of breakpoints in the Structural Break competition. From the description, it’s not entirely clear what the exact semantics of the break point and the period=1 label are. I’d appreciate your clarification on the following points:
a) Is the labeled break point (i.e., the index where period=1 begins) always intended to mark the “start” of the new data-generating process?
Or in other words: Does the label y=True & period=1 imply that this sample — and all samples after — are generated from the new (post-break) process?
b) Alternatively, could there be situations where the actual change in the data-generating process occurs a few steps before or after the indicated break point?
For example, if y=False at the indicated breakpoint, does that mean some of the samples in period=1 may have still been generated by the pre-break process?
Or, is it False because the post-break process had actually started before the labeled breakpoint?
Understanding this would help clarify whether “period=1” can be treated as a reliable marker for the transition in distribution, or if it could lag behind the actual structural change.
Thanks in advance!