ETA accuracy benchmark · 🇬🇷 Piraeus

How Port Flow's predicted ETA compares to each vessel's own broadcast ETA at Piraeus, 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)

17.40 h

MAE 7.01 h · 961 voyages

Broadcast ETA (RMSE)

67.85 h

crew-declared, 366 voyages (same set)

Our advantage

74.4 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (389 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)

MMSI	Cargo	Arrival	Our error	Broadcast error
636021664	container	07/12, 11:57 AM UTC	2.55 h	-2.96 h
240205400	product	07/12, 11:29 AM UTC	0.20 h	614.27 h
240043800	product	07/12, 10:50 AM UTC	0.08 h	—
255805908	container	07/12, 10:43 AM UTC	1.00 h	-837.72 h
240708000	chemical	07/12, 09:13 AM UTC	1.56 h	-77.23 h
241750000	product	07/12, 09:12 AM UTC	2.31 h	-917.21 h
253000158	container	07/12, 08:39 AM UTC	0.39 h	-363.66 h
240205400	product	07/12, 07:37 AM UTC	3.44 h	618.14 h
240698300	container	07/12, 07:12 AM UTC	1.74 h	—
240698300	container	07/12, 06:57 AM UTC	0.49 h	-29.46 h
247485200	container	07/12, 06:55 AM UTC	2.32 h	-373.92 h
240620300	product	07/12, 06:45 AM UTC	1.18 h	2478.23 h
241694000	container	07/12, 05:02 AM UTC	2.42 h	-0.55 h
241694000	container	07/12, 04:39 AM UTC	1.09 h	-23.17 h
241842000	container	07/12, 01:21 AM UTC	1.22 h	-0.28 h
241324000	product	07/12, 12:22 AM UTC	2.23 h	3298.62 h
249427000	general-cargo	07/11, 09:27 PM UTC	1.05 h	-90.46 h
352006400	container	07/11, 08:57 PM UTC	1.06 h	-126.96 h
210989000	container	07/11, 08:25 PM UTC	1.58 h	-102.42 h
240205400	product	07/11, 08:05 PM UTC	0.50 h	629.67 h
240620300	product	07/11, 07:19 PM UTC	1.95 h	2489.68 h
247104500	product	07/11, 07:16 PM UTC	2.41 h	-682.28 h
240620300	product	07/11, 06:48 PM UTC	-3.77 h	2490.19 h
271051219	general-cargo	07/11, 05:29 PM UTC	1.99 h	-254.49 h
538007627	product	07/11, 05:02 PM UTC	4.80 h	-726.04 h
240897000	product	07/11, 04:57 PM UTC	0.92 h	4159.03 h
636020667	container	07/11, 04:03 PM UTC	1.61 h	-230.06 h
255805788	container	07/11, 03:11 PM UTC	2.81 h	-246.18 h
240205400	product	07/11, 02:05 PM UTC	3.30 h	635.67 h
247305000	container	07/11, 01:54 PM UTC	0.53 h	-292.92 h
240183900	product	07/11, 12:42 PM UTC	2.22 h	-50.71 h
255805996	container	07/11, 11:11 AM UTC	1.16 h	-209.20 h
240897000	product	07/11, 10:50 AM UTC	1.04 h	4165.15 h
538010615	product	07/11, 10:45 AM UTC	1.63 h	-239.75 h
240752600	product	07/11, 10:42 AM UTC	1.40 h	—
239786000	container	07/11, 10:26 AM UTC	1.70 h	-36.44 h
237829800	product	07/11, 10:14 AM UTC	1.19 h	—
565671000	container	07/11, 09:35 AM UTC	2.40 h	-6.60 h
241671000	container	07/11, 08:18 AM UTC	2.07 h	-34.55 h
237592400	product	07/11, 07:01 AM UTC	-45.06 h	—
241654000	product	07/11, 06:59 AM UTC	2.70 h	3334.00 h
239869900	product	07/11, 06:59 AM UTC	0.53 h	4156.51 h
237006500	container	07/11, 06:52 AM UTC	4.40 h	-33.87 h
305473000	container	07/11, 06:47 AM UTC	0.65 h	-208.80 h
220129000	container	07/11, 06:22 AM UTC	1.05 h	-881.37 h
636023590	container	07/11, 05:42 AM UTC	1.69 h	-3.70 h
240732000	product	07/11, 05:14 AM UTC	3.90 h	-49.24 h
240732000	product	07/11, 04:24 AM UTC	3.15 h	-144.41 h
240043800	product	07/11, 01:47 AM UTC	1.20 h	—
240949000	product	07/11, 12:58 AM UTC	3.43 h	-7.98 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.