Abandonment differs from drift in being abrupt. There is a moment - almost always identifiable afterwards - when the trader stopped consulting the plan and started reacting. Everything after it belongs to a different strategy with no tested properties.
The cost is not only the losses in that stretch. Those trades contaminate the record, so the next review mixes tested behaviour with improvisation and neither can be evaluated.
Mark it when it happens. A single flag in the journal for trades taken off-plan keeps the two populations separate, and the comparison between them is usually persuasive enough to reduce the behaviour on its own.
Related: rule-drift, tilt, behaviour-journal, forced-trading