ETA accuracy benchmark · 🇬🇧 Milford Haven
How Port Flow's predicted ETA compares to each vessel's own broadcast ETA at Milford Haven, measured on closed voyages over the last 30 days. We publish the number — good or bad. Lower error is better.
Port Flow predicted ETA (RMSE)
52.82 h
MAE 25.89 h · 16 voyages
Broadcast ETA (RMSE)
Building…
only 8 comparable voyages so far (need 20)
Gap
—
benchmark builds as more voyages close
Closed voyages
error = predicted/broadcast minus actual arrival (hours)| MMSI | Cargo | Arrival | Our error | Broadcast error |
|---|---|---|---|---|
| 244066000 | product | 07/12, 03:01 AM UTC | -78.17 h | — |
| 244046000 | product | 07/08, 06:43 PM UTC | 3.83 h | -22.73 h |
| 255915968 | product | 07/08, 03:32 PM UTC | 5.38 h | -1003.54 h |
| 241963000 | product | 07/08, 02:18 PM UTC | 7.50 h | -92.30 h |
| 244056000 | product | 07/08, 11:55 AM UTC | -8.29 h | -882.93 h |
| 352006501 | product | 07/08, 08:41 AM UTC | -34.16 h | -107.69 h |
| 636021956 | product | 07/08, 01:26 AM UTC | -1.15 h | — |
| 244066000 | product | 07/07, 01:32 AM UTC | -5.05 h | -20.05 h |
| 538007646 | product | 07/06, 06:08 AM UTC | -18.69 h | -26.14 h |
| 241445000 | product | 07/02, 06:53 PM UTC | -10.89 h | — |
| 249224000 | product | 06/28, 05:17 AM UTC | -15.17 h | -12.29 h |
| 241445000 | product | 06/26, 12:40 AM UTC | -34.81 h | -89.67 h |
| 255806263 | product | 06/25, 01:34 AM UTC | 0.82 h | — |
| 244066000 | product | 06/24, 11:47 PM UTC | -2.40 h | -132.80 h |
| 249224000 | product | 06/20, 07:29 AM UTC | 0.22 h | — |
| 232064564 | product | 06/17, 07:02 PM UTC | -187.76 h | — |
See it on a live desk
Predicted ETAs, congestion and sanctions screening across 51 ports — try it in one click, no signup.
How it's measured
- Predicted ETA: distance-to-berth over speed-made-good, refreshed on every AIS position, with a seasonal correction from rolling medians.
- Broadcast ETA: the crew-entered ETA in the vessel's AIS message — the de-facto industry default.
- Both are compared to the actual arrival timestamp on the same closed voyages. We report RMSE (penalises large misses) and MAE (typical error).
- Broadcast ETAs more than 7 days off are dropped as AIS sentinel / placeholder values, and our model is scored on that same set — so the comparison is honest, not cherry-picked.
- Coverage is public AIS only; ports with weak coverage show fewer closed voyages. We'd rather show a small honest sample than a big misleading one.