ETA accuracy benchmark · 🇧🇷 Santos

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

14.68 h

MAE 10.32 h · 522 voyages

Broadcast ETA (RMSE)

61.35 h

crew-declared, 253 voyages (same set)

Our advantage

76.1 %

lower error than the broadcast ETA

Broadcast ETAs more than 7 days from the actual arrival are excluded as AIS sentinel / placeholder values (162 excluded). Both figures are computed on the same remaining voyage set — apples to apples.

Closed voyages

error = predicted/broadcast minus actual arrival (hours)


712204006	product	10/12, 01:54 AM UTC	-0.31 h	—
710004636	general-cargo	10/11, 11:47 PM UTC	0.33 h	-624.79 h
257199000	product	10/11, 11:40 PM UTC	1.98 h	-34.67 h
219505000	container	10/11, 11:13 PM UTC	2.26 h	-336.72 h
249163000	product	10/11, 10:51 PM UTC	0.91 h	-609.36 h
563758000	container	10/11, 10:47 PM UTC	-0.48 h	-244.79 h
710100009	product	10/11, 10:05 PM UTC	0.54 h	—
229553000	product	10/11, 06:44 PM UTC	2.27 h	-27.73 h
712204006	product	10/11, 04:59 PM UTC	-0.29 h	—
712204006	product	10/11, 04:41 PM UTC	-0.38 h	—
352003659	product	10/11, 04:32 PM UTC	1.03 h	-28.54 h
538008595	container	10/11, 01:31 PM UTC	0.38 h	-133.53 h
414665000	container	10/11, 12:39 PM UTC	-0.05 h	-0.66 h
477174900	container	10/11, 09:16 AM UTC	1.09 h	616.73 h
257524000	container	10/11, 09:11 AM UTC	1.48 h	-4.19 h
710100009	product	10/11, 08:48 AM UTC	-0.14 h	—
538003692	container	10/11, 08:46 AM UTC	3.78 h	-149.28 h
566794000	container	10/11, 05:18 AM UTC	0.33 h	-0.82 h
352003837	container	10/11, 04:43 AM UTC	1.45 h	-181.73 h
477113100	general-cargo	10/11, 04:34 AM UTC	0.52 h	-164.58 h
712204006	product	10/11, 03:38 AM UTC	1.05 h	—
710100009	product	10/11, 01:39 AM UTC	1.58 h	—
636025634	container	10/11, 12:19 AM UTC	1.05 h	-11.33 h
710071000	container	10/11, 12:08 AM UTC	0.11 h	-132.13 h
538010606	container	10/11, 12:02 AM UTC	0.55 h	-335.04 h
636025568	container	10/10, 11:13 PM UTC	1.93 h	-178.23 h
636026671	container	10/10, 11:06 PM UTC	-0.01 h	-1.10 h
636023145	container	10/10, 11:04 PM UTC	0.45 h	-1.57 h
710100009	product	10/10, 09:32 PM UTC	-0.49 h	—
219107000	container	10/10, 06:07 PM UTC	1.04 h	14.87 h
636018721	container	10/10, 05:39 PM UTC	0.69 h	-294.66 h
636022122	container	10/10, 05:38 PM UTC	-0.23 h	-137.64 h
255916520	chemical	10/10, 05:17 PM UTC	0.51 h	-3.80 h
563033800	container	10/10, 05:13 PM UTC	0.39 h	-25.73 h
538009615	product	10/10, 02:12 PM UTC	2.19 h	-214.21 h
710100009	product	10/10, 01:37 PM UTC	-0.19 h	—
712204006	product	10/10, 08:04 AM UTC	0.90 h	—
712204006	product	10/10, 07:42 AM UTC	2.99 h	—
712204006	product	10/10, 07:14 AM UTC	-0.18 h	—
247341900	container	10/10, 07:09 AM UTC	0.32 h	-31.16 h
710100009	product	10/10, 06:34 AM UTC	1.04 h	—
538011110	container	10/10, 05:31 AM UTC	2.65 h	-314.52 h
371956000	container	10/10, 05:15 AM UTC	0.30 h	-16.25 h
710006782	general-cargo	10/10, 05:08 AM UTC	0.89 h	-39.14 h
538009665	container	10/10, 05:07 AM UTC	0.35 h	-62.12 h
249397000	container	10/10, 04:52 AM UTC	0.85 h	-28.88 h
255917223	dry-bulk	10/10, 01:17 AM UTC	0.38 h	-37.29 h
710100009	product	10/10, 12:09 AM UTC	0.14 h	—
712204006	product	10/09, 11:41 PM UTC	0.01 h	—
477155900	container	10/09, 11:27 PM UTC	0.34 h	-4.21 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.