ETA accuracy benchmark · 🇫🇷 Le Havre

How Port Flow's predicted ETA compares to each vessel's own broadcast ETA at Le Havre, measured on closed voyages over the last 90 days. We publish the number — good or bad. Lower error is better.

Port Flow predicted ETA (RMSE)

4.41 h

MAE 1.10 h · 987 voyages

Broadcast ETA (RMSE)

57.53 h

crew-declared, 463 voyages (same set)

Our advantage

92.3 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (313 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)


245889000	product	10/12, 04:33 AM UTC	-0.14 h	20.44 h
232046826	general-cargo	10/12, 03:29 AM UTC	0.36 h	-262.49 h
353083000	container	10/11, 11:39 PM UTC	0.47 h	-10.16 h
205250990	container	10/11, 11:09 PM UTC	0.04 h	—
353083000	container	10/11, 10:45 PM UTC	-0.58 h	-795.75 h
205250990	container	10/11, 10:16 PM UTC	0.68 h	—
259739000	container	10/11, 10:14 PM UTC	0.80 h	45.76 h
636025289	product	10/11, 09:39 PM UTC	1.13 h	-6.65 h
219031111	product	10/11, 09:34 PM UTC	-1.56 h	-9.58 h
259739000	container	10/11, 09:28 PM UTC	1.58 h	46.52 h
255805557	container	10/11, 08:57 PM UTC	-0.35 h	-51.96 h
235104739	chemical	10/11, 06:57 PM UTC	0.29 h	16.05 h
226005390	product	10/11, 06:34 PM UTC	1.58 h	-406.09 h
265012000	container	10/11, 05:00 PM UTC	1.39 h	-2.52 h
226005390	product	10/11, 03:52 PM UTC	-0.14 h	-403.39 h
265012000	container	10/11, 03:50 PM UTC	0.61 h	-1.34 h
235096859	product	10/11, 03:31 PM UTC	2.56 h	-11.52 h
226009040	product	10/11, 03:24 PM UTC	0.48 h	5861.75 h
226009040	product	10/11, 02:46 PM UTC	0.42 h	5862.38 h
235096859	product	10/11, 02:20 PM UTC	1.01 h	-78.35 h
215746000	chemical	10/11, 01:01 PM UTC	2.14 h	-14.03 h
215746000	chemical	10/11, 12:18 PM UTC	1.26 h	-498.32 h
228437800	product	10/11, 11:56 AM UTC	4.04 h	0.05 h
636018716	product	10/11, 11:27 AM UTC	-0.58 h	-146.46 h
255805810	container	10/11, 11:12 AM UTC	0.56 h	-2506.21 h
311001583	product	10/11, 10:15 AM UTC	-0.49 h	-7.26 h
636093166	general-cargo	10/11, 08:54 AM UTC	0.23 h	-1.91 h
255801480	container	10/11, 08:39 AM UTC	0.14 h	-1.66 h
226017260	chemical	10/11, 08:34 AM UTC	1.38 h	7210.42 h
636093166	general-cargo	10/11, 08:22 AM UTC	-0.14 h	-1.38 h
226017260	chemical	10/11, 08:13 AM UTC	2.22 h	7210.77 h
255801480	container	10/11, 07:56 AM UTC	0.64 h	-142.95 h
228499700	container	10/11, 07:46 AM UTC	1.02 h	-2285.78 h
226005390	product	10/11, 07:37 AM UTC	0.57 h	-395.14 h
255803590	container	10/11, 05:11 AM UTC	-0.53 h	-611.19 h
636022611	container	10/11, 04:10 AM UTC	0.24 h	-182.17 h
256220000	product	10/11, 01:34 AM UTC	0.11 h	-4.58 h
245889000	product	10/10, 11:18 PM UTC	2.56 h	-32.78 h
319445000	product	10/10, 10:31 PM UTC	3.17 h	-12.53 h
245889000	product	10/10, 10:17 PM UTC	-0.17 h	-74.30 h
226021210	container	10/10, 09:52 PM UTC	0.58 h	—
226021210	container	10/10, 09:07 PM UTC	1.09 h	-350.12 h
319445000	product	10/10, 08:56 PM UTC	1.25 h	-10.94 h
228437800	product	10/10, 07:40 PM UTC	2.47 h	—
228437800	product	10/10, 06:48 PM UTC	0.48 h	-4.81 h
228437800	product	10/10, 06:31 PM UTC	0.03 h	-230.53 h
259739000	container	10/10, 06:22 PM UTC	0.99 h	-2.88 h
259739000	container	10/10, 05:17 PM UTC	0.18 h	-1.79 h
259739000	container	10/10, 04:48 PM UTC	-0.19 h	-1.30 h
226009040	product	10/10, 04:34 PM UTC	0.89 h	—

See it on a live desk

Predicted ETAs, congestion and sanctions screening across 51 ports — try it in one click, no signup.

Open dashboard See pricing →

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.