ETA accuracy benchmark · 🇺🇸 New York / Newark

How Port Flow's predicted ETA compares to each vessel's own broadcast ETA at New York / Newark, measured on closed voyages over the last 7 days. We publish the number — good or bad. Lower error is better.

Port Flow predicted ETA (RMSE)

9.46 h

MAE 3.87 h · 108 voyages

Broadcast ETA (RMSE)

33.84 h

crew-declared, 33 voyages (same set)

Our advantage

72.0 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (35 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)


432666000	container	10/12, 04:11 AM UTC	2.25 h	-3.20 h
441344000	container	10/12, 01:58 AM UTC	-0.63 h	-6.47 h
636019342	general-cargo	10/12, 01:51 AM UTC	-1.97 h	—
338590000	product	10/12, 12:25 AM UTC	-45.92 h	-160.43 h
538012806	container	10/11, 11:05 PM UTC	0.55 h	-3.09 h
366999412	container	10/11, 06:31 PM UTC	-0.34 h	—
636092927	container	10/11, 03:16 PM UTC	2.63 h	-564.28 h
255915817	container	10/11, 12:13 PM UTC	1.19 h	—
636025655	container	10/11, 11:26 AM UTC	0.70 h	-2.19 h
212175000	container	10/11, 10:30 AM UTC	2.90 h	-2230.52 h
258758000	container	10/11, 10:24 AM UTC	9.09 h	716.59 h
636019342	container	10/11, 09:44 AM UTC	9.34 h	-1274.75 h
563051200	container	10/11, 09:34 AM UTC	2.51 h	-1029.57 h
563190500	container	10/11, 07:01 AM UTC	0.43 h	-2107.02 h
271052065	container	10/11, 06:00 AM UTC	1.70 h	-4.51 h
566956000	container	10/11, 03:10 AM UTC	0.15 h	-932.67 h
357961000	container	10/10, 11:38 PM UTC	2.98 h	-0.64 h
305969000	general-cargo	10/10, 10:12 PM UTC	0.04 h	-773.20 h
477222700	general-cargo	10/10, 07:46 PM UTC	2.21 h	-1693.78 h
366999412	container	10/10, 06:52 PM UTC	-0.50 h	—
255915817	container	10/10, 12:12 PM UTC	0.72 h	-3.70 h
367359930	product	10/10, 12:05 PM UTC	0.00 h	1810.90 h
255915619	container	10/10, 12:01 PM UTC	0.46 h	-2530.02 h
366999412	container	10/10, 11:46 AM UTC	-17.22 h	—
247341700	container	10/10, 11:44 AM UTC	16.33 h	-1491.74 h
215800000	container	10/10, 08:58 AM UTC	-1.87 h	1.03 h
367361960	product	10/10, 08:24 AM UTC	-15.65 h	-1747.41 h
563293100	container	10/10, 02:01 AM UTC	1.92 h	-1887.03 h
370651000	container	10/10, 01:49 AM UTC	0.21 h	-1902.83 h
303657000	container	10/09, 11:33 PM UTC	2.11 h	—
636019145	product	10/09, 11:28 PM UTC	0.50 h	-3.99 h
366999412	container	10/09, 05:54 PM UTC	-0.03 h	—
366999412	container	10/09, 05:48 PM UTC	-0.59 h	—
215800000	container	10/09, 03:59 PM UTC	0.63 h	-836.99 h
563020400	general-cargo	10/09, 01:23 PM UTC	0.61 h	-2056.39 h
366999412	container	10/09, 11:26 AM UTC	-16.89 h	—
367065790	product	10/09, 10:18 AM UTC	-0.89 h	—
477648800	container	10/09, 09:50 AM UTC	1.82 h	-3014.84 h
563231100	container	10/09, 08:55 AM UTC	-0.07 h	-937.93 h
367686920	product	10/09, 06:09 AM UTC	-15.19 h	—
636016363	chemical	10/09, 06:01 AM UTC	0.22 h	-24.03 h
538009014	product	10/09, 01:28 AM UTC	-20.26 h	-19.48 h
357961000	container	10/08, 10:47 PM UTC	0.70 h	-410.79 h
538008204	container	10/08, 09:48 PM UTC	2.14 h	-83.80 h
255806376	container	10/08, 08:53 PM UTC	1.07 h	-3.39 h
310764000	container	10/08, 08:51 PM UTC	0.24 h	-1183.85 h
215577000	container	10/08, 07:01 PM UTC	-0.70 h	-6.77 h
366999412	container	10/08, 05:59 PM UTC	-0.19 h	—
367359930	product	10/08, 03:33 PM UTC	-1.97 h	1855.44 h
255806034	container	10/08, 03:23 PM UTC	1.87 h	-3.90 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.