What sixteen platforms look like from the inside.
Reading every major competition platform once a day produces a second dataset alongside the listings: a record of how the platforms themselves behave. How late they publish. How often they answer. What they quietly stop returning. None of that is published anywhere else, so it is published here, with the method stated and the numbers recomputed rather than maintained.
Study 01
How much warning do students actually get?
Every listing is stamped the day it appears, so the gap between appearing and closing is measurable per platform. This is the study that says whether checking a site on Sunday is a strategy or a gamble.
Method: Deadline minus first-seen date, per platform, from the daily archive.
Study 02
Which competition platforms can be relied on?
Sixteen feeds, read on a schedule, with every outcome recorded. Uptime, response time and quiet failure per platform — including our own, since a source of ours that breaks belongs on a public page.
Method: Every scheduled run since the archive began, grouped by source.
Using any of this?
Everything here is free to quote with a link back. The numbers move as the archive grows, so cite the page rather than copying a figure, and if you need the underlying rows the listings are available as an open API. Journalists and researchers wanting a cut we do not publish can ask.