Calling all users of Trackernet/tube predictions

Do you use Trackernet or live tube predictions in our Unified API? If so we’d love to know:

  • If you use the Trackernet legacy feeds or the Unified API
  • What you use the data for in your products
  • What could be improved in our tube predictions data.

Please spill the beans on this thread!

Thank you :grinning:


I coded for the Unified API, and then had to re-write as the only Trackernet can do start-of-line departures which includes the Jubilee at Stratford and Central Line at Ealing Broadway.

The unified api needs to include departures information. Arrivals is no good at termini or if mode is terminating.

Unified api also needs to include stopping pattern information for tube, eg, fast, semi fast, all stations trains as seen on the Met.

Also would be good to see mainline trains (non tfl) returned within the api. The legacy api is partly capable of this but calling a destination “TOC” is not helpful enough.

Many tube calls in the unified api show duplicated trains. These should be removed with only one instance remaining as some locations can return 2-3 instances of the same train.

1 Like

Ideally tube predictions would provide the same information that you can see on the displays in the stations, but they currently don’t do this. I’ve noticed the following differences:

  • The predictions provide arrival times for trains already on their journey. This means no departure times for terminal stations and very limited departure information for other stations near the end of the line.
  • They don’t always provide the platform number for stations that have multiple platforms for trains traveling in the same direction (eg Acton Town).
  • They include duplicate trains.
  • There can be noticeable latency between the API data and station departure boards.

To be honest, the current data isn’t really that useful as it’s not accurate or complete enough to use it for planning a journey. It’s considerably less useful than the departure times provided by national rail.

In an ideal world low latency predictions would be available via push technology (eg Serer-Sent Events or WebSockets in a way that doesn’t require a propitiatory third party library) so that it is possible to create departure boards that are in sync with those on the platform.

@briantist How are you able to get start-of-line departures from Trackernet? If I look at data for Ealing Broadway then I can see DepartTime attribute for trains at the platform, but it appears to be the time the train arrived at the platform.

1 Like


I think, on reflection, that I made something that was “good enough” (it showed something, which the staff found useful) rather than the Unfiied API which shows nothing.

The whole Ealing Broadway to North Acton thing is a bit of a oddity. Quite why they turn trains back at White City rather than doing the step-back at Ealing Broadway I don’t know, it sees a very poor use of the link that Ealing Broadway has to Tfl Rail/Elizabeth Line.