Prediction accuracy

How accurate are Wolfline arrival times?

Updated Jun 20, 12:00 PM EDT · refreshed hourly

84%

land inside the accuracy window

scored by the open Transit/Swiftly ETA benchmark

6×

more accurate than the schedule

near arrival, 21s error vs 123s

96%

within one minute as it arrives

when your bus is 0–3 minutes away

✓18,667 predictions checked·7 routes·last 30 days

How accurate, by how soon your bus arrives

Each column is a time-to-arrival window from the open Transit/Swiftly benchmark. The green band is the accepted accuracy window: it narrows the closer your bus is, because that's when the number matters most. The closer the bus, the more of our predictions land inside it.

catch the ride (accurate)excess wait (bus late)miss the ride (bus early)

Each dot ≈ 88 predictions · 18,667 total · overall accuracy is the straight average of the four buckets: 84%

Prediction accuracy by time-to-arrival window, scored by the Transit/Swiftly ETA benchmark. Overall accuracy is 84% across 18,667 predictions.
Time to arrival	Accuracy	Predictions
10 to 15 minutes away	85% accurate	4,532
6 to 10 minutes away	79% accurate	4,624
3 to 6 minutes away	84% accurate	4,710
0 to 3 minutes away	90% accurate	4,801

Accuracy over time

Daily, since measurement began.

Daily prediction accuracy from Jun 19, 2026 to Jun 20, 2026.
Date	Accuracy	Predictions
Jun 19, 2026	85% accurate	15,504
Jun 20, 2026	81% accurate	3,163

When we're off, which way?

A bus arriving earlier than predicted is the one a rider misses, so we watch that closely.

on target

7% late

8.7% early

When a prediction misses, a bus running early (8.7%) is the costly kind: you can't catch one that's already left, while running late (7%) just means a brief wait.

Accuracy by route (3–6 min out)

Route	Checks	Median error	Within 2 min
52	1,183	52s	88%
41	1,145	50s	87%
30	1,037	39s	91%
42	424	50s	81%
43	402	56s	77%
60	317	57s	84%
20	202	48s	93%

How we measure this

Every time a bus arrives, we look back at the prediction we were showing minutes earlier, sampled across the trip and bucketed by time-to-arrival exactly as the open-source Transit/Swiftly ETA Accuracy Benchmark specifies (0–3, 3–6, 6–10 and 10–15 minutes out). A prediction counts as accurate if it landed inside that bucket's asymmetric window, stricter on early arrivals, because a bus that has already left is one you can't catch. The headline figure is the benchmark's straight average of the four buckets. Built entirely from raw GPS, refreshed hourly.

Wolfline · measurement window since 2026-06-19