Confirmation probability answers one question.
“Will this waitlisted ticket confirm?” is a question about a ticket. Journey Success asks something wider: “Across this complete plan, what share of simulated outcomes end with me completing the journey?”
What gets simulated?
Every candidate journey is run through 10,000 seeded simulations. Each run draws a ticket outcome and a delay for every leg from what we know about that train, then checks whether the connections survived and whether you arrived in time. Journey Success is simply the share of those runs that got you there.
Seeded means repeatable: the same journey and the same evidence produce the same number, so a figure you saw an hour ago has not quietly drifted for no reason.
Can you board each leg?
Known availability and confirmation estimates are used where we have them. Where we do not, that run stays explicitly low-confidence rather than optimistic.
How late does each leg arrive?
Delays are drawn from how the train has actually been running, instead of pretending every train keeps exactly to the timetable.
Does each connection survive?
If the arrival delay eats the usable transfer buffer, that run fails at the connection — which is exactly what would have happened to you.
Do you reach in time?
If you told us when you have to be there, a journey that otherwise completed can still fail for arriving too late.
The number stays explainable.
Journey Success always comes with the reasons it is not higher: how many runs failed on the ticket, how many on a missed connection, how many on arriving late. “68%” on its own tells you to worry. “68%, and almost all of the failures are the connection at DDU” tells you what to do about it.
It is also kept apart from the Recommendation Score. Journey Success is a probability — a fact about the world. The Recommendation Score is about you: whether you asked to reach safely, to get there fast or to spend less. Telling us what matters changes the ranking. It never changes the probability.
A simulation is not current truth.
We label a simulation as a simulation. A current seat count, a confirmation estimate and a Journey Success figure are three different kinds of thing, and presenting them as interchangeable facts would be the easiest way to make this product look confident and be wrong.