Do Last-Minute Openings Vanish Faster?

Every cancellation hunter carries a theory of the clock. The most common one goes like this: the closer a trip gets, the more frantic the market — so an opening that surfaces in the final days must vanish in minutes, while an opening that surfaces weeks ahead can sit there, comparatively unhurried, for the patient. It is a plausible theory. It shapes real behavior: people set frantic refresh loops for this weekend's trip and casual ones for next month's. We can now test it, because our engine does not just detect openings — it follows each one until two consecutive polls confirm it is gone, and logs the full life cycle. The sample for this post is the 274 completed life cycles recorded between October 30 and 31, 2026, for stays in November and December 2026.

Survival by lead time: the flat clock

Lead time is the gap between the day an opening was detected and the first night of the stay. Split the 274 life cycles into buckets and take the median survival of each:

Opening appeared	Completed life cycles (n)	Median survival
2–7 days ahead	148	1h 52m
8–14 days ahead	90	1h 49m
15+ days ahead	36	2h 38m

Read that table again, because it is not the table intuition promised. The 2–7 day bucket — the heart of the supposedly frantic final week — lives a median 1h 52m. The 8–14 day bucket lives 1h 49m — statistically the same clock. Only the 15+ day bucket runs visibly slower, at 2h 38m, and even that is a difference of tens of minutes, not of kind. There is no bucket in this data where an opening lives long enough to be caught by an evening check, and no bucket where it dies so fast that only machines need apply. The machine-vs-human gap we documented at Upper Pines applies at every lead time, almost uniformly.

Why would the clock be flat? Because the two forces intuition separates actually scale together. Yes, a near-date opening faces more desperate hunters — but it also faces more supply: the same deadline pressure that makes hunters frantic makes holders decide, so near-date openings are born more often (our days-ahead measurement put the median appearance at days, not weeks, before the stay). More hunters chasing more openings clears at roughly the same rate as fewer hunters chasing fewer. The market's temperature rises; its clearing speed, apparently, does not.

The variable that does move the clock: stay length

Split the same 274 life cycles by how many nights the opening offered, and a real gradient appears:

Opening length	Completed life cycles (n)	Median survival
1 night	134	1h 59m
2 nights	64	1h 37m
3+ nights	76	1h 13m

Openings of three or more nights die fastest — a median 1h 13m, against 1h 59m for single nights. The mechanism is not mysterious: multi-night openings are the scarce, universally usable inventory. A single night fits only hunters whose plans can absorb a fragment; a three-night block fits almost everyone's plan exactly as drawn, so the queue for it is the whole market. Scarcity plus universal fit equals speed. It is also why the single-night fragments that dominate the flow (166 of the desert openings in our companion analysis were single nights; the market-wide pattern is the same) are the patient hunter's opportunity: they live longest precisely because they are hardest to use.

What this changes, practically

Do not triage your attention by lead time. The trip in five days and the trip in twenty-five days deserve the same response speed — about 1h 51m at the median — because the data shows the same clock on both. If anything, the far-out opening deserves more respect than intuition gives it: a 15+ day opening living 2h 38m is not a "check tonight" opening either. Do triage by stay length. When a 3+ night opening at your target appears, that is the fastest clock in the dataset — decide in minutes, with dates and acceptable sites settled beforehand. When single nights appear, you have comparatively more room — use it to stitch fragments into a full stay rather than waiting for a perfect block that, on this evidence, will be gone in 1h 13m when it finally shows up. Keep the watch running at every distance. A flat clock means there is no "too early to matter" and no "too late to bother": the fortnight and the month-out window run on the same physics.

The limits, plainly

Three, stated as loudly as the finding. First, the missing bucket: the current closure log contains no completed life cycles with a lead under 2 days, so this post says nothing — in either direction — about openings for tonight or tomorrow; they may well run on a different clock, and we will publish it when the sample exists. Second, right-censoring: openings still alive when the sample closed are excluded, biasing every median here downward, most for the far-out bucket whose openings have had the least time to die — the true 15+ day median is somewhat slower than printed. Third, season: this is a shoulder-season sample (November–December stays). A July sample could steepen every gradient on this page. The live counts are on the cancellation index and the method is on the methodology page; we will republish this split as the log grows. Until then, plan on the flat clock: at every lead time we can measure, you have about two hours, and for the openings worth most, closer to one.

Frequently asked questions

Last updated 2026-10-30.