Try to think of this in terms of how you will use the model. It sounds like a day-trading model, correct me if I’m wrong. In this case, you’ll want to ask the question of “based on todays trading patterns, should I sell now, or is the peak still likely to come?”.

See if this helps your problem formulation and therefore your labeling.

As a side note, most models are not sophisticated enough to capture the extreme complexity of stock behavior. If this is your first foray into stock prediction, I’d recommend tempering expectations.