Live proof

THOR publishes its measured accuracy, airport by airport and horizon by horizon (2 h, 1 h, 30 min and 15 min before arrival), against actual landing times. A figure is published only from 30 measured flights in the segment: below that, only the number of flights is shown.

Loading measurements…
AirportHorizonPeriodFlightsMean errorP90BiasWithin ±10 min

Method

  • Measurement points are chosen by time before arrival, never by error.
  • Flights without a landing time and collection gaps stay in the denominators; nothing is removed afterwards.
  • The engine and its calibration are locked by fingerprint; any change is dated.

Historical cohort (P2)

On the forward cohort of 95 flights (0–2 h), THOR’s mean error was 7.98 minutes against 14.95 minutes for the reference, and THOR did better on 65.3% of flights.