yes, good point. we base our statistical inferences on backward-looking data, but the future can change in ways we're unable to fully predict and account for.

this illustrates that just-in-time isn't an unalloyed good. it trades off resiliency for cost (and a bit of agility). it also introduces more structural variance in the supply chain, exacerbating supply issues, which isn't obvious on first blush (operations research coursework covers this).