Voraussichtliche Zielzeiten anhand von Kontrollpunkten beim London-Marathon 2026, wenn die Läufer ihr aktuelles Tempo bis zum Ziel beibehalten. Scharfe Spitzen bei den üblichen Rundenzielzeiten (3:00, 3:30, 4:00) glätten sich im Laufe des Rennens.
Von Last_Kick9059
10 Kommentare
Hate that you cant pause this
Looks like a lot of people are aiming to get under 4 hours, but then reality sets in as they get tired.
What might be a fascinating follow-up would be some way to visualize how many people from the inital bucket ended up in each eventual bucket.
**Data source:** [https://results.tcslondonmarathon.com/2026/](https://results.tcslondonmarathon.com/2026/)
**Tools used:** beautiful soup + python
The problem seems to be that those are individual projections aggregated instead of a normal distribution. So if someone reaches 5km in X minutes, the projection says that he will reach the marathon distance in Y minutes and that time is always the same. A projection would get much more accurate if they used historic data to get a distribution instead of a fixed Y for every X.
The projected time should have the caveat of „if all goes well“, which it seems it doesn’t go well about 1/4 of the time
My goal: 3h55m
Actual time: 4h22m
Graph checks out!
The peaks e.g on the hour are interesting. Almost like a bunch of people who would otherwise be slightly slower push themselves to break a particular hour boundary
Worth noting that the first 5k is all downhill so can be slightly faster in the first 5k than their planned overall race pace rather than this being pacing gone wrong
This is great! Would you mind sharing where you got the data from please? Is it freely available?
Are those spikes just before 3 and 4 hours real or an artifact of the data?