ETA accuracy benchmark · 🇺🇸 Houston

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

12.49 h

MAE 7.69 h · 172 voyages

Broadcast ETA (RMSE)

73.96 h

crew-declared, 63 voyages (same set)

Our advantage

83.1 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (90 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)


232032457	container	10/12, 05:20 AM UTC	12.23 h	-7.34 h
636024112	product	10/12, 04:59 AM UTC	-0.36 h	-652.98 h
256721000	product	10/12, 04:23 AM UTC	-33.92 h	-252.40 h
563022800	product	10/12, 04:03 AM UTC	-1.69 h	-1890.06 h
565516000	product	10/12, 01:49 AM UTC	-2.95 h	-1041.83 h
538009778	chemical	10/12, 01:18 AM UTC	2.39 h	-2907.31 h
351298000	container	10/12, 01:03 AM UTC	8.65 h	-7.05 h
249198000	chemical	10/12, 01:00 AM UTC	3.37 h	-1014.51 h
255915818	container	10/11, 11:44 PM UTC	2.75 h	-4.23 h
405000346	container	10/11, 09:08 PM UTC	3.42 h	-11.14 h
636013521	container	10/11, 05:47 PM UTC	-1.80 h	-3.80 h
352005659	container	10/11, 05:34 PM UTC	1.25 h	—
636017165	chemical	10/11, 03:45 PM UTC	3.41 h	-196.76 h
563027500	product	10/11, 02:26 PM UTC	4.56 h	-1087.44 h
304824000	container	10/11, 01:57 PM UTC	4.50 h	-2977.45 h
255806138	product	10/11, 12:40 PM UTC	-20.40 h	-151.68 h
255916495	product	10/11, 11:46 AM UTC	-0.10 h	-574.78 h
477124300	container	10/11, 11:28 AM UTC	-0.35 h	-86.48 h
636015526	product	10/11, 10:13 AM UTC	3.33 h	-24.22 h
255803260	container	10/11, 09:54 AM UTC	-0.69 h	-4.91 h
636018471	container	10/11, 09:43 AM UTC	-3.13 h	-2898.72 h
248411000	container	10/11, 07:20 AM UTC	-0.25 h	-4.34 h
538007042	product	10/11, 05:14 AM UTC	8.52 h	-8.74 h
636018714	container	10/11, 05:02 AM UTC	0.67 h	-163.04 h
636017417	product	10/11, 04:36 AM UTC	8.62 h	-4.61 h
241408000	product	10/11, 02:15 AM UTC	-1.42 h	-297.26 h
538004411	container	10/11, 12:06 AM UTC	-2.76 h	-204.11 h
563093400	product	10/10, 08:27 PM UTC	6.16 h	-38.46 h
241387000	product	10/10, 07:13 PM UTC	-13.85 h	-43.25 h
538010978	product	10/10, 06:54 PM UTC	10.33 h	-457.91 h
538004912	general-cargo	10/10, 04:49 PM UTC	-1.71 h	-3.82 h
367775000	container	10/10, 10:37 AM UTC	0.36 h	-3.13 h
311001673	product	10/10, 10:27 AM UTC	3.56 h	-1460.46 h
412763000	container	10/10, 10:25 AM UTC	-18.71 h	-23.43 h
256916000	product	10/10, 10:14 AM UTC	-3.08 h	-2713.24 h
257453000	product	10/10, 06:48 AM UTC	-15.54 h	-24.31 h
367497070	product	10/10, 06:14 AM UTC	10.82 h	5822.29 h
368611000	container	10/10, 06:12 AM UTC	-8.05 h	—
538005911	product	10/10, 05:59 AM UTC	1.68 h	-2908.99 h
538011999	container	10/10, 05:57 AM UTC	4.30 h	-24.95 h
563048700	chemical	10/10, 05:45 AM UTC	7.95 h	-118.75 h
241348000	product	10/10, 05:08 AM UTC	-15.25 h	-3147.14 h
431230000	product	10/10, 04:51 AM UTC	0.22 h	-3189.86 h
636017950	product	10/10, 04:19 AM UTC	12.22 h	-2498.32 h
538009780	chemical	10/10, 03:17 AM UTC	7.03 h	-81.29 h
352003540	product	10/10, 03:08 AM UTC	4.34 h	-1255.14 h
353423000	product	10/10, 02:09 AM UTC	8.19 h	-729.16 h
636012425	product	10/10, 02:03 AM UTC	-46.70 h	-59.56 h
636021492	container	10/10, 12:11 AM UTC	-6.33 h	-608.19 h
563272400	product	10/09, 10:41 PM UTC	-0.75 h	-29.68 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.