ETA accuracy benchmark · 🇮🇹 Genoa

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

8.47 h

MAE 13.19 h · 987 voyages

Broadcast ETA (RMSE)

78.97 h

crew-declared, 363 voyages (same set)

Our advantage

89.3 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (456 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)


215325000	container	10/12, 05:18 AM UTC	0.49 h	-1.01 h
236509000	container	10/12, 05:13 AM UTC	2.00 h	-781.23 h
215940000	container	10/12, 05:08 AM UTC	-0.20 h	-1788.14 h
563183100	container	10/12, 04:50 AM UTC	-0.41 h	-48.84 h
247453600	general-cargo	10/12, 04:22 AM UTC	1.70 h	-1.37 h
247453600	general-cargo	10/12, 04:10 AM UTC	0.32 h	-114.93 h
247304200	container	10/11, 07:57 PM UTC	0.68 h	-1198.96 h
249319000	container	10/11, 05:48 PM UTC	1.58 h	-2.80 h
247380600	container	10/11, 03:26 PM UTC	-0.13 h	7011.56 h
247287200	product	10/11, 03:20 PM UTC	-0.21 h	-305.59 h
247380600	container	10/11, 02:04 PM UTC	0.61 h	7012.92 h
255805929	container	10/11, 01:54 PM UTC	0.62 h	-0.41 h
247388500	container	10/11, 01:31 PM UTC	0.36 h	-458.03 h
247287200	product	10/11, 01:27 PM UTC	1.38 h	—
255805834	container	10/11, 01:20 PM UTC	0.40 h	-74.84 h
247270900	product	10/11, 10:10 AM UTC	0.01 h	7161.82 h
247388500	container	10/11, 09:03 AM UTC	-0.31 h	-453.56 h
247287200	product	10/11, 07:24 AM UTC	-0.47 h	—
310826000	container	10/11, 06:56 AM UTC	1.53 h	-1113.93 h
247380600	container	10/11, 06:54 AM UTC	0.22 h	7020.08 h
256070000	container	10/11, 05:44 AM UTC	-0.03 h	-141.24 h
205703000	chemical	10/10, 06:00 PM UTC	1.90 h	—
205703000	chemical	10/10, 05:40 PM UTC	-0.93 h	-18.67 h
636023127	general-cargo	10/10, 04:35 PM UTC	0.84 h	-451.59 h
247270900	product	10/10, 04:22 PM UTC	0.24 h	—
247061700	container	10/10, 03:53 PM UTC	0.85 h	1.11 h
247287200	product	10/10, 02:36 PM UTC	-0.22 h	-280.87 h
247270900	product	10/10, 02:29 PM UTC	0.91 h	7181.51 h
247388500	container	10/10, 02:18 PM UTC	-0.45 h	-434.80 h
247287200	product	10/10, 11:08 AM UTC	1.65 h	—
247287200	product	10/10, 11:04 AM UTC	0.32 h	—
247270900	product	10/10, 10:51 AM UTC	0.75 h	—
255916531	container	10/10, 10:13 AM UTC	-0.31 h	-51.23 h
247388500	container	10/10, 09:00 AM UTC	-0.23 h	-429.50 h
247270900	product	10/10, 08:45 AM UTC	0.02 h	—
255806391	container	10/10, 08:29 AM UTC	-0.07 h	-1.49 h
229296000	general-cargo	10/10, 08:15 AM UTC	0.86 h	-411.26 h
255806056	container	10/10, 07:54 AM UTC	1.08 h	-0.90 h
255806056	general-cargo	10/10, 07:44 AM UTC	0.22 h	-99.73 h
255916593	container	10/10, 07:29 AM UTC	-0.16 h	-63.49 h
247380600	container	10/10, 07:27 AM UTC	1.22 h	7043.55 h
215548000	container	10/10, 07:04 AM UTC	-0.72 h	-70.07 h
247270900	product	10/10, 06:47 AM UTC	1.32 h	7189.21 h
247380600	container	10/10, 06:37 AM UTC	0.92 h	7044.38 h
248717000	container	10/10, 06:35 AM UTC	0.51 h	-145.59 h
247383400	container	10/10, 06:03 AM UTC	1.48 h	-65.56 h
357627000	container	10/10, 05:43 AM UTC	0.13 h	-215.73 h
247297200	container	10/10, 04:34 AM UTC	-0.24 h	-0.82 h
247297200	container	10/10, 04:23 AM UTC	0.19 h	-66.64 h
240630000	container	10/10, 04:05 AM UTC	0.28 h	-534.09 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.