Every travel-savings app asks you for one number and hands back one date. Both are false. A trip does not cost one number — flights, lodging, food and the exchange rate are each a range, and some of them move together. And the day you can go is not one day: it depends on what you actually manage to set aside, which is rarely what you planned.
Wayfare reports an earliest and a latest month, with the most likely month drawn faintly between them. The visual hierarchy is inverted on purpose: in this category the single big number is the headline and the uncertainty is a footnote. Here the two bounds are the headline, because they are the part that is actually known.
If the trip runs two nights longer, lodging and food and local transport all rise at once. Adding them as though they were unrelated shows a range narrower than the truth. Wayfare asks which items share a cause — the number of nights, the number of people, the exchange rate — and adds those as one thing.
In an ordinary three-item example the naive sum looks 1.37× narrower than the honest one. The app shows you that number rather than describing it.
How long until you have enough is a ratio: what is still missing, divided by what you set aside each month. Both are uncertain. When the second is uncertain enough — a few uneven months — the honest answer has no upper end at all. The interval comes from Fieller’s theorem, which is built for ratios and can return an unbounded answer. The usual shortcut, the delta method, cannot: it quietly prints a finite month, and it is wrong.
“Nine times out of ten” is easy to write. The small-sample correction inside Wayfare was not assumed — it was found by simulation, then re-measured on data it had never seen, with a seed and a grid of parameters that were never used in fitting:
| Complete months | Unevenness | Range held |
|---|---|---|
| 4 | 0.18 | 90.9% |
| 4 | 0.62 | 90.8% |
| 6 | 0.38 | 90.4% |
| 12 | 0.38 | 90.5% |
| 48 | 0.18 | 93.7% |
That measurement is a file in the source, and the build fails if the worst cell ever falls below the promise. An earlier version of the correction, fitted to exactly 90%, measured 89.7% on a fresh seed — a promise tuned to the boundary is kept half the time. The target was moved above the boundary and the whole table was re-measured.
You note the rate yourself, whenever you happen to see it. With fewer than four notes Wayfare says the rate is not in the range yet, rather than quietly assuming a currency does not move. That is the entire reason the app needs no network at all — and the reason nothing about where you are going, or when, ever reaches anyone.
Each cost as a range: at least this, at most that. If you truly know a figure, a visa fee say, enter it once and its range is zero wide. Then your deposits, as they happen. There is nothing to connect and nothing to sign in to.
Only complete calendar months count, and months with nothing in them count as zero. Both rules exist because the alternatives are silently wrong: a half-finished month drags the rate down, and skipping empty months nearly doubles it.
The same two dates the phone worked out. The watch computes nothing of its own — two devices measuring the same trip would give two different answers, and that would end your trust in both.
Wayfare makes no network request of any kind: not for rates, not for analytics, not for anything. No bank connection, no account, no sign-in, no sync, no server, no ads. It asks for no permissions — no contacts, location, notifications or camera. Each of those lines is a build gate that fails the build if it stops being true — see Privacy.
Wayfare reports an interval computed from numbers you typed in, and never recommends an action, a product, a provider or an amount. It is not affiliated with any bank, broker or lender.
Free. No in-app purchases, no subscription. Everything runs on your device, so there is nothing to bill for.
iPhone, iPad and Apple Watch. English, Turkish, German, French, Spanish and Japanese.
To wayfare is to travel, and a wayfarer is someone on the road with a distance still to cover. This one tells you how far off the far end really is.