Replication of The Sailing as a Scheduled Value Coverage: Preregistration and Code Freeze

Preregistered plan and frozen analysis code for a replication of Guarino (2026), The Sailing as a Scheduled Value Coverage: Dimensionless Purchase Evaluation for Multi-Port Itineraries (doi:10.5281/zenodo.23077791). The August 2026 study's 83-sailing dataset, plan and scripts were not kept, so its printed figures cannot be reproduced. This replication draws a new random sample of 83 sailings (40 Caribbean, 19 European, 24 Alaska and Pacific) from a stated frame, builds the scheduled value coverage by the paper's rules, and reruns its three tests with the original thresholds: H1, whether the construction restates cost per night (redundant if |rho| >= 0.90); H2, whether fare per night buys scheduled shore time (passes if the 95 percent interval contains zero); and H3, whether the itinerary groups improve holdout prediction of fare (passes if the median gain exceeds 0.02). The plan was written on 1 October 2026, before any replication data was collected. Its SHA-256 is a6e7052fad547120dec8f37985c0f033faeb123680281e7cfc590ecfc98c70a9, and that hash is also the seed of the random draw, so the sample was fixed when the plan was. The code freeze holds the construction and test scripts with their hashes, and a dry run on labelled synthetic data. The real data will be run through exactly these files. Any later departure from either will be recorded in a dated deviations log and reported with the results.