And it will happily tell you train X departed from Y when in reality you're in that train and stuck at two stations before Y. I have a logger running, but I'm pretty sure the data will turn out useless as I've noticed this was not unusual to happen. Note: the trains have GPS and you can follow them in the official NS API (or if you reverse the app, or use a site that did that).
NDOV is much more up to date, though not perfect either.
In short: no API gives you actual truth based on GPS or something°, and the NS API itself is unreliable and unmaintained (their app is way beyond what is provided to the public).
° you could say this is nearly impossible, but one could use their GPS and align trains to a track within 15 meters of the reported location, then determine whether it's on schedule. Train tracks are pretty much perfect in OSM data (and I think there are official sources in some weird format from the 80's it seems, designed by HP of all companies) so that's a fine source. There are also sites that interpolate expected position based on schedule and delay data with surprising accuracy (50 meters on a good day I'd estimate, or 500m or so on average), so it's not hard to compare expected vs. actual and determine a delay.
Source: worked on rpln.nl since they killed their mobile site in favor of 5 megabytes of "responsive javascript". Came across a lot of stuff while doing that, and we're still working on features and feature parity with the official stuff (we're feature compatible with the old mobile site, even bookmarks work if you just update the domain, but not as far as the official site because of the unmaintained api).
The UK NR API is pretty awesome. It is a STOMP feed of every train passing location. So it is pretty much exactly accurate, as the train physically has to pass the reporting sensor for it to update. If the train hasn't passed that sensor, it doesn't move.
While not 15m accuracy it's pretty damn good. Probably <500m in urban areas and ~1-2km in rural areas.
NDOV is much more up to date, though not perfect either.
In short: no API gives you actual truth based on GPS or something°, and the NS API itself is unreliable and unmaintained (their app is way beyond what is provided to the public).
° you could say this is nearly impossible, but one could use their GPS and align trains to a track within 15 meters of the reported location, then determine whether it's on schedule. Train tracks are pretty much perfect in OSM data (and I think there are official sources in some weird format from the 80's it seems, designed by HP of all companies) so that's a fine source. There are also sites that interpolate expected position based on schedule and delay data with surprising accuracy (50 meters on a good day I'd estimate, or 500m or so on average), so it's not hard to compare expected vs. actual and determine a delay.
Source: worked on rpln.nl since they killed their mobile site in favor of 5 megabytes of "responsive javascript". Came across a lot of stuff while doing that, and we're still working on features and feature parity with the official stuff (we're feature compatible with the old mobile site, even bookmarks work if you just update the domain, but not as far as the official site because of the unmaintained api).