From 2026-11-04 to 2026-11-18 · computed 2026-11-18 04:30 UTC
How accurate Trenalia and its sources are, measured against what happened afterwards: every night, what the site showed is compared with the real arrival, measured by Trenalia from station boards and from the GPS position recorded throughout the journey. It is fully automatic: nothing is entered by hand.
11.1 minmean error of Trenalia's delay (official: 10.4 min)
38.1 %samples within ±3 min of the real arrival delay
297journeys with an observed arrival (36,201 samples)
66.8 %network alerts confirmed (4,356 checked)
1 · What happened at the end of the journey
Trenalia measures on its own when each train arrived: from the destination station's arrivals board (ADIF or RadarDeTrenes) when available, otherwise from the continuously recorded GPS position (entering within 700 m of the station, or an estimate from the last position and speed if it vanishes within 3 km). That arrival is compared here with Renfe's last published delay. Positive bias = Renfe reports more delay than observed.
Samples
Mean difference (min)
Agree within ±2 min
Renfe bias (min)
p90
All observed arrivals
297
5.1
53.9 %
2.5
12.4
Destination arrivals board
231
5.5
49.4 %
3.5
13
GPS: entering the station
53
3.9
77.4 %
-0.9
6.8
GPS: estimated when signal lost nearby
13
2.4
38.5 %
-2.4
4
Boards as the train passes each station
26,127
3.6
65.5 %
1.4
8
2 · Delay shown vs. observed arrival
The toughest test: what the site showed at each moment is compared with the arrival measured by Trenalia (board or GPS), not with Renfe's figure. “Official” is Renfe's published delay; “Trenalia” the projected one. Gain = how much Trenalia reduces the official error.
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
Total
36,201
10.4
11.1
44.9 %
38.1 %
-0.6
-6.4 %
0–15 min before arrival
3,682
5.3
5.9
59.6 %
54.8 %
4.1
-10.9 %
15–30 min before arrival
3,902
6.7
7.5
53.5 %
47.6 %
3.4
-11.9 %
30–60 min before arrival
7,233
7.6
8.3
50.3 %
44.1 %
1.9
-9.8 %
60–120 min before arrival
10,569
10.7
11.5
39.6 %
33.3 %
0
-7.5 %
2–4 h before arrival
10,814
15.2
15.6
38.3 %
29.5 %
-5.8
-3 %
By service
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
Media Distancia and regional
17,866
14.4
14.8
39.3 %
33.8 %
-4.2
-3 %
AVE
8,148
4.3
5.1
62.7 %
52.9 %
2.7
-18.2 %
Alvia
3,615
13.1
13.7
31 %
27.2 %
-0.6
-4.5 %
Avant
2,837
3.2
4.1
67.3 %
55.6 %
2.3
-27.6 %
Euromed and Intercity
2,382
7.7
8.6
30.4 %
26.7 %
6.1
-12 %
Avlo
1,353
8.1
10.3
27.3 %
16.7 %
10.3
-27 %
By day
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
2026-11-04
3,190
7.2
7.7
54.5 %
46.9 %
-0.1
-7.6 %
2026-11-05
1,377
8.2
8.2
51.9 %
45.9 %
-1.4
-0.1 %
2026-11-06
3,505
8
8.5
46.1 %
39.8 %
2.5
-6.1 %
2026-11-07
2,156
9.5
9.5
36.3 %
33.3 %
-2.5
-0.3 %
2026-11-08
3,848
6.2
7
45.5 %
37.4 %
2.7
-13.4 %
2026-11-09
2,604
9
10.5
41.3 %
30.6 %
4.6
-16.4 %
2026-11-10
474
11.3
9.2
32.5 %
33.3 %
-5.1
+18.3 %
2026-11-11
1,959
13.4
15.2
41.8 %
27 %
-2.6
-13.9 %
2026-11-12
2,599
36.8
38.3
49.4 %
42.4 %
-23.4
-4.1 %
2026-11-13
3,446
9
8.8
47.4 %
42.9 %
-2.6
+2.2 %
2026-11-14
3,078
6.2
7.3
48.3 %
40.5 %
3
-17.1 %
2026-11-15
1,664
8.3
10.4
39.8 %
31.4 %
10
-25.4 %
2026-11-16
3,969
9.3
9.2
42.1 %
35.1 %
1.1
+1.7 %
2026-11-17
2,332
8.3
8.9
36.7 %
37.3 %
1.1
-7.4 %
3 · Delay shown vs. Renfe's last figure
The same, taking as final the last delay Renfe published at the end of the journey (available for almost every journey, but not an independent measure). It shows whether Trenalia anticipates what Renfe will end up reporting.
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
Total
42,709
9.4
10.1
44.8 %
41.3 %
-1.7
-7.5 %
0–15 min before arrival
4,206
1.4
2.6
90.5 %
79.3 %
2.1
-85.9 %
15–30 min before arrival
4,403
3.4
4.3
71.2 %
59.9 %
2.6
-26.9 %
30–60 min before arrival
8,345
6.3
6.6
53.6 %
50.1 %
0.4
-5 %
60–120 min before arrival
12,381
10.5
11
34.7 %
32.8 %
-1.9
-4.5 %
2–4 h before arrival
13,373
14.9
15.9
25.6 %
25.7 %
-5.5
-6.3 %
By service
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
Media Distancia and regional
23,555
12.2
13.3
39.5 %
37.2 %
-3.8
-9.1 %
AVE
8,401
4.5
4.7
57.7 %
53.9 %
0.6
-3.9 %
Alvia
4,076
10.3
9.9
35 %
31.3 %
-3.6
+4.7 %
Avant
2,873
3.5
4.2
65.2 %
57.1 %
2.6
-20 %
Euromed and Intercity
2,410
8.1
9.2
34.7 %
31.1 %
7.4
-12.8 %
Avlo
1,394
4.7
4.7
61 %
50.1 %
0.1
+0.5 %
By day
Samples
Official error (min)
Trenalia error (min)
±3 min official
±3 min Trenalia
Bias (min)
Gain
2026-11-04
3,822
6.9
6.6
54.7 %
49.3 %
-2.7
+5.1 %
2026-11-05
2,169
7.1
7.2
49.8 %
44.9 %
-1.2
-2.2 %
2026-11-06
4,505
8.3
8.6
42.6 %
39.8 %
-0.4
-3.3 %
2026-11-07
2,781
8.5
11.3
41.2 %
35.9 %
-0.3
-33.7 %
2026-11-08
4,779
7.6
7.4
42.6 %
40.8 %
-1.7
+2.1 %
2026-11-09
2,903
10
11.1
35.3 %
31.4 %
3
-11.2 %
2026-11-10
762
7.3
14.6
48.6 %
37.4 %
8.4
-99.9 %
2026-11-11
2,243
7.2
8
35.8 %
36.2 %
-0.1
-11.8 %
2026-11-12
2,720
33.5
34.1
54.7 %
49.2 %
-27.4
-1.8 %
2026-11-13
3,652
6.3
8.3
53 %
44.1 %
2
-31.5 %
2026-11-14
3,789
7.4
9.2
42.4 %
38.7 %
3.3
-24.6 %
2026-11-15
1,942
5.8
5.9
44.9 %
42.4 %
-0.1
-1.1 %
2026-11-16
4,012
9.3
8.1
44 %
43.6 %
-2.5
+12.5 %
2026-11-17
2,630
8.8
9.1
37.5 %
40 %
0.2
-3.5 %
4 · Learned models
Trained every night on the history and evaluated on the last week, which is held out from training. They are only switched on if they clearly beat the baseline.
Delay correction (ETA)
rejected no mejora lo bastante a la referencia · trained 2026-11-18 04:30 UTC
Model mean error (min)
2.348
Baseline mean error (min)
2.272
Gain (fraction)
-0.034
R²
0.038
Training examples
97,034
Validation examples
52,966
Trees
37
Validation from
2026-11-12
False alarm filter for closing alerts
rejected AUC insuficiente · trained 2026-11-18 04:30 UTC
AUC
0.732
Training examples
3,260
Validation examples
816
False alarm share
0.918
Alerts that would be hidden (p ≥ 0.9)
709
Precision when hiding
0.989
Trees
35
5 · Train closing alerts
There is no official data to compare them with, so we label what happened afterwards. If the gap kept closing, the alert was consistent with the data; if the train ahead was stopped at a station, it was probably a false alarm (normal operations).
Candidates evaluated
2,378
Alerts shown
2,044
Still closing 2 min later
11 %
Likely false alarm (train ahead at a station)
0.5 %
What happened afterwards
Automatic check 3 h after each alert shown: whether both trains finished their journey normally, whether Renfe or Adif published an incident in the area, and whether there were network alerts nearby. “No signs” is not official confirmation: official accident data is published by the CIAF.
No signs
973
Incident in the area
579
Mentioning collision or derailment
0
Signs to review
492
No data
0
Reviewed by hand
0
6 · Network alerts
A segment or stopped-train alert counts as confirmed if any train involved gained 3 min or more of delay 10 to 30 min later; a sudden delay jump, if it did not revert (it was not a data glitch).
Alerts
Checked
Confirmed
Train stopped outside a station
3,699
3,335
58.1 %
Sudden delay jump
806
784
98.7 %
Segment with several trains losing time
240
237
84.4 %
Total
4,745
4,356
66.8 %
7 · Data quality
Days with trajectories
15
Position samples per day
102,671
Positions snapped to the track
99.1 %
Estimated precision (median)
189 m · p90 987 m
8 · Incident hypotheses
Trenalia opens a hypothesis when it sees, live, a stopped train or several trains in trouble on a section, and checks it by itself 3 h 30 min later against the actual delays. Every night it also re-simulates the previous day's situations with the engine to see whether it reproduces them. They are not Adif or Renfe notices.
Hypotheses checked live
Samples
Confirmed
With official notice
Passenger votes
Line closure or interruption
755
77.7 %
59
0
Stopped or broken-down train
2,702
78.2 %
339
0
Re-simulation of past situations
Samples
Reproduced
Line closure or interruption
4,793
0 %
Stopped or broken-down train
33
6.1 %
Slow running
169
3 %
Learned patterns: 1,713 · validated by the maintainer: 82 · proposed: 0 · retired: 317
What Trenalia tells travellers
Accuracy of the arrival times people see in “I’m on a train” and “Your train” (last 14 days): share of times the time shown was within ±5 min, by how long before arrival and the confidence given. Cases in brackets.
High confidence
Medium confidence
Low confidence
60 min before arrival
50 % (8)
—
33 % (3)
30 min before arrival
17 % (6)
—
100 % (1)
15 min before arrival
17 % (6)
—
100 % (1)
5 min before arrival
0 % (3)
—
—
Arrived more than 15 min later than stated
0 %
—
20 %
Only trains and stations are counted (all queries for the same train and station are merged), never people. The actual arrival is the last estimate in its final minutes or the moment the train was seen passing.