Our track record
Every launch cycle, this engine either calls the window or stays silent. Both are on the record. Here is every call to date — how early, from what evidence, and what actually happened.
Warning time, flight by flight
View as table
Bar length = days between the window first becoming publicly knowable and liftoff. Test events (static fires) are excluded — they have no launch to lead.
Starship Flight 11
CALLED ITALERT — window confirmed Oct 9, 2025 · 4 days before liftoff
Outcome: Launched on the primary date we identified.
Backtest replay: our engine, run against archived copies of the closure page exactly as it appeared in 2025, fires the booking alert on Oct 9 naming Oct 13 as the primary date.
Starship Flight 12
CALLED ITMarine hazard window detected May 20, 2026 · 2 days before liftoff
Outcome: Launched inside the Coast Guard hazard window.
Reconstructed from the USCG District 8 broadcast notice (BNM 0055-26) our marine-notice source now ingests; the county closure page had gone stale and missed this flight entirely — the reason our engine watches six sources instead of one.
Starship Flight 13
CALLED ITHazard windows July 23–29 published July 20, 2026 · 3 days before first attempt
Outcome: Aborted July 16, weather-scrubbed July 23, launched July 24 — inside the published backup-day sequence.
Reconstructed from USCG BNM 0080-26, which listed seven consecutive daily launch windows including every backup day. A subscriber booking to cover that window would have been on the island for the launch.
Static-fire campaign (test, not a launch)
CORRECTLY SILENTNO ALERT — correctly identified as a test
Outcome: Closures posted, slipped twice, and cleared with no launch. Anyone treating closures alone as a launch signal would have booked a hotel for an engine test.
Live: our engine tracked all three closure revisions in real time and held at "verifying" because the dates did not match the public launch window. Not alerting is the hard half of this job.
Honesty notes
The chart covers every Starship flight ever flown. Flights 1–10 are reconstructed from Internet Archive snapshots of the closure page; where the archive has gaps, we use the latest possible first-signal date, so those lead times are conservative minimums — the real warning may have been longer. Flights 11–13 predate this site's launch (August 30, 2026), so those entries are backtests and reconstructions: our engine run against archived copies of the public pages as they existed at the time, and official Coast Guard records — not predictions we published in advance. The static-fire entry is live tracking. From Flight 14 onward, every entry is a timestamped public call made before the outcome. We publish misses the same way we publish hits.
Alerts we got wrong
Every message that went to subscribers and should not have, or should have read differently. Keeping this list is the point of the promise above.
Sent 95 minutes after a successful launch, telling subscribers the primary date had moved to Sep 29. The flight was already over.
Cause: Same frozen calendar cache: the tracker still believed the flight was ahead of it. · fixed 2026-09-28
Sent 75 minutes AFTER the rocket had lifted off, and listed Sep 29 as both a primary and a backup date.
Cause: The calendar cache had never expired, so nothing could tell us the flight had launched; and the city page carries two closure blocks that disagree after a scrub. · fixed 2026-09-28
We told subscribers the attempt was delayed and the closures withdrawn, the night before the launch. It was false: the closures were never withdrawn and the launch flew on schedule.
Cause: A single failed fetch of the city page emptied that cycle’s closure list, and the engine read an unreadable page as an empty one. · fixed 2026-09-28
We said "new primary Sep 28" when the primary had not moved at all — a backup day had been added. It read as though the launch date had changed.
Cause: One update template assumed any change meant the primary moved. · fixed 2026-09-26
Our heads-up said the launch was likely the week of Sep 22. It flew Sep 28. We never sent a correction, because the date-moved notice needs manual approval and none was given.
Cause: Heads-up fires on a community target date; the target slipped twice afterwards.
The next entry on this page will be Flight 15 — currently targeting December 2026. Our call will appear here, timestamped, before the launch — subscribe to get it as an alert instead of reading it afterward.