The finding
Time to departure predicted the size of a move better than its direction.
At 31–60 days out, 18.0% of verified weekly pairs fell—essentially the same as 18.3% at 121+ days. The move itself was different.
The 31–60 day result covers 1,722 comparisons across 196 sailings and 20 lines. The final 14 days are still provisional: only 67 comparisons met the same continuity test.
Booking-window curve
Most exact fares held for a week. Among movers, the balance stayed mixed.
Each row answers: if this exact cabin fare was available at both weekly checkpoints, did its advertised base fare go down, stay flat, or go up?
Move severity
Both the upside and downside widen near departure.
This is why a last-minute strategy is a wager, not a rule. At 15–30 days, the average observed cut was -12.4%, while the average increase was +17.1%.
In this sample, waiting did not make a cut more likely. It increased the stakes when a price moved. If your fare rules allow repricing, booking an acceptable fare and continuing to watch preserves more options than waiting for a pattern that may break either way.
Cruise-line sample
The market average hides very different pricing rhythms.
This table uses all booking windows in the study. “Down” and “up” count any week-over-week base-fare change, not only moves above a marketing threshold. Small samples should be treated as directional.
Three lines fell more often than they rose.
- Norwegian43.4% down · 22.1% up
- Princess49.5% down · 18.3% up
- Royal Caribbean48.3% down · 27.6% up
Carnival moved up far more often.
Disney and Ritz-Carlton usually held.
88.8% of Disney pairs and 93.7% of Ritz-Carlton pairs were unchanged. Their rarer moves could still be large.
31–60 days · 249 Disney and 191 Ritz-Carlton pairsThese callouts use one comparable window. Royal Caribbean has the smallest sample here—58 pairs across six sailings—so treat its direction as provisional.
| Cruise line | Selected sailings | Verified pairs | Down | Up | Avg. move |
|---|---|---|---|---|---|
| Atlas | 22 | 116 | 19.8% | 2.6% | 5.5% |
| Azamara | 60 | 1,821 | 26.6% | 31.7% | 11.4% |
| Carnival | 60 | 2,244 | 29.6% | 47.4% | 2.1% |
| Celebrity | 60 | 1,341 | 36.6% | 45.5% | 9.2% |
| Celebrity Expeditions | 7 | 31 | 16.1% | 32.3% | 10.5% |
| Costa | 60 | 1,670 | 11.4% | 10.0% | 11.3% |
| Crystal | 42 | 716 | 0.8% | 14.1% | 9.6% |
| Cunard | 60 | 1,795 | 17.3% | 16.3% | 10.5% |
| Disney | 60 | 2,064 | 3.7% | 13.4% | 10.7% |
| Explora | 60 | 1,427 | 17.3% | 12.0% | 8.2% |
| Four Seasons | 30 | 512 | 1.8% | 3.3% | 24.6% |
| Holland America | 60 | 2,305 | 17.7% | 8.6% | 5.4% |
| MSC | 60 | 1,912 | 11.9% | 20.0% | 7.9% |
| Norwegian | 60 | 2,730 | 36.8% | 14.7% | 5.7% |
| Oceania | 60 | 941 | 6.8% | 20.9% | 4.6% |
| Princess | 60 | 1,768 | 35.0% | 41.1% | 5.8% |
| Ritz-Carlton | 60 | 1,981 | 0.6% | 1.2% | 16.6% |
| Royal Caribbean | 60 | 2,243 | 28.3% | 33.4% | 7.1% |
| Seabourn | 60 | 922 | 9.5% | 17.7% | 5.7% |
| Silversea | 60 | 878 | 13.8% | 21.6% | 6.3% |
| Viking | 60 | 1,213 | 10.1% | 10.3% | 6.0% |
| Viking Expeditions | 15 | 301 | 7.6% | 3.7% | 3.3% |
| Virgin | 60 | 2,328 | 16.7% | 19.4% | 12.5% |
Methodology
The same fares were tracked over time, instead of using a rotating group of the cheapest cabins.
1,196 currently bookable sailings across all 23 tracked lines, capped at 60 deterministically selected sailings per line. Where fewer qualified, we used all of them.
USD advertised base fare per person at double occupancy, for the same sailing and exact cabin category exactly seven days apart.
Only the current, uninterrupted availability history was retained. A cabin category that disappeared and returned did not bridge the missing period.
Observed asking prices are not booked prices. We do not observe remaining inventory, demand, or whether a traveler could reprice an existing reservation.
Coverage: 1,292,844 stored price states across 206,857 double-occupancy fare series, collected Jul 7, 2026–Sep 24, 2026. The published sample is intentionally smaller because continuity and exact-match filters are strict. Market curves pool verified category-level pairs, so lines with more categories contribute more comparisons; the line table exposes that distribution. The report can be regenerated from the same query as new daily observations arrive.