FOI into the calculation of National Rail fares

[Nick W]

I recently submitted a FOI request to DfT asking for information the process involved in calculating an individual railway fare dynamically when it was requested. It seems that they were rather naive about the handling of large amounts of data and the processes that we can logically infer are involved, and denied that such a process exist.

Those of you who are interested in the request and the response, can find it here.

 

[Peter Mugridge]

Interesting that you have chosen to quote that particular flow as an example for the request - I am actually doing that specific trip very soon as a charity fundraising event!


And yes, I agree there must be something more to it than the dfT are trying to say.

 

[yorkie]

The TOCs decide the fares, and they are done as 'flows'.

Penzance to Wick is priced however XC and East Coast choose to price it.

I could have saved a bit of public money and your time by telling you that Nick

Interestingly they provided a little bit of info about ORCATS (but nothing we already knew I don't think?) which determines how the revenue is allocated, I think this missed the point you were asking which is how the ticket price is calculated, which you are never going to be able to get them to answer.

If you want to know how any particular TOC decides to rip us off (in most cases) with the fares, I suggest you write to that TOC, but the chances of them actually giving a meaningful answer are negligible.

 

[Nick W]

Looking back, my maths was totally wrong. For some reason I thought they'd be number of stations factorial tickets, when actually it's an arithmetic series.

If however, we assume that there's about 1000 key stations on the network (a rough assumption), there's still half a million possible combinations looking at one type of ticket (eg anytime return) alone. If we assume that just the one ticket type alone takes a person 10 minutes to decide a good fare, it'd take a year! For this reason, at least for 90% of fares, I'm confident that automated methods exist for pricing the respective fares.

I can however conclude that fares between key stations are stored individually on the ATOC fares cd, with other stations referring to the individual key stations.

If you want to know how any particular TOC decides to rip us off (in most cases) with the fares, I suggest you write to that TOC, but the chances of them actually giving a meaningful answer are negligible.

I would have thought fares such as Wick - Penzance would be determined by ATOC in consultation with the respective TOCs, rather than XC choosing by its own method.

 

[yorkie]

Looking back, my maths was totally wrong. For some reason I thought they'd be number of stations factorial tickets, when actually it's an arithmetic series.

If however, we assume that there's about 1000 key stations on the network (a rough assumption), there's still half a million possible combinations looking at one type of ticket (eg anytime return) alone. If we assume that just the one ticket type alone takes a person 10 minutes to decide a good fare, it'd take a year! For this reason, at least for 90% of fares, I'm confident that automated methods exist for pricing the respective fares.

Erm, yes, usually RPI + 1%. With today's computers it probably takes less than a second

I can however conclude that fares between key stations are stored individually on the ATOC fares cd, with other stations referring to the individual key stations.

I'm not sure what you mean or how you can draw conclusions but storing thousands of fares in a database and performing a function on them just isn't a problem these days.

I would have thought fares such as Wick - Penzance would be determined by ATOC in consultation with the respective TOCs, rather than XC choosing by its own method.

Why would that fare not be set by XC and EC (for the applicable routeings)? What's different about Wick to Penzance?

There do appear to be some fares set by ATOC but they are for a specific reason, e.g. in the London Zones, where it is probably TfL and ATOC liaising to determine the fares, as they are zonal, and TOCs cannot exempt themselves from these zonal fares. I seem to recall there may have been some legislation passed a few years ago to do this.

But, other than that, the TOCs decide.

 

[Mojo]

By derivation; do you mean what used to appear in the National Fares Manual book for some stations? In which case the fare for Penzance to Wick is derived from Penzance to Thurso.

 

[Nick W]

By derivation; do you mean what used to appear in the National Fares Manual book for some stations? In which case the fare for Penzance to Wick is derived from Penzance to Thurso.

To clarify, this would be the first step - what I'd really like to know is how the fare from Penzance to Thurso itself was calculated.

Yorkie has probably hit the nail on the head - most non-local fares have probably only risen by around RPI + 1% since the BR days, so it would really be a question of how BR calculated their fares.

I'm not sure what you mean or how you can draw conclusions but storing thousands of fares in a database and performing a function on them just isn't a problem these days.

Indeed, I foolishly believed at first that the number of journey opportunities between say 5 stations was 1x2x3x4 rather than 1+2+3+4, and therefore ruled storing all the fares in a database impractical. Turns out it is much for practical than I had first believed.

 

[tony_mac]

Erm, yes, usually RPI + 1%. With today's computers it probably takes less than a second

Yes, but the prices must have come from somewhere! and they should be reviewed occasionally.

If the fares are reviewed, there must be some method of working with the data.
There are 2,500 stations - so 3+ million pairs of stations, with sometimes many ticket types available isn't really practical. (and some fares are not bi-directional!)

There are 209 routeing points, so still 20,000+ pairs to work with - and some fares are definitely not related to routeing points.

Although, if you could have any number of routes for each ticket, then it would, indeed, be a factorial number of fares. (and you would need very large tickets as well )

But, I would guess it's just an ad-hoc process; they look at a pair they want to consider (i.e. increase!), and then check the nearby ones to make sure they are already high enough and increase them if not!

I'm not sure what you mean or how you can draw conclusions but storing thousands of fares in a database and performing a function on them just isn't a problem these days.

That doesn't make much sense to me either, an index to a key station would take about the same amount of space as a storing the fare.

Why would that fare not be set by XC and EC (for the applicable routeings)? What's different about Wick to Penzance?

The whole idea that a single company 'controls' a ticket price that many (or almost all) other operators have to accept on their trains is a bit crazy!

 

[yorkie]

Yorkie has probably hit the nail on the head - most non-local fares have probably only risen by around RPI + 1% since the BR days.

Well, that was an example really, but that's what happens to regulated fares at the moment. (did it used to be RPI -1% in the early days under Labour?)

But the point is that any TOC can just put all their fares up by a certain percentage, and that's probably what they generally do.

However some TOCs have put them up higher than others, and this is one reason why anomalies have crept in!

The whole idea that a single company 'controls' a ticket price that many (or almost all) other operators have to accept on their trains is a bit crazy!

Indeed it is crazy!

By the way, the DfT really don't like FOI requests, although this one gave them an easy way out as it's flawed. They do realise that some of the FOI requests originate from people on this forum and they do read it!

 

[Paul Kelly]

Hello there, I've found the RJIS Data Feed Specification for Fares and Associated Data to give some useful hints about how this sort of thing works. The diagram on page 8 shows a block "Files from Fares Maintenance System" as an input into the RJIS data factory - perhaps the "Fares Maintenance System" is only hypothetical though! Also interesting is the distinction it makes between "derivable" and "non-derivable" fares. The fact that some fares are derivable suggests to me that there is some kind of algorithm whereby every single combination of point-to-point fares does not need to be stored in the database. The document does not appear to give any detail on how the derivable fares may be derived though...

 

[barrykas]

Also interesting is the distinction it makes between "derivable" and "non-derivable" fares. The fact that some fares are derivable suggests to me that there is some kind of algorithm whereby every single combination of point-to-point fares does not need to be stored in the database. The document does not appear to give any detail on how the derivable fares may be derived though...

Presumably a similar way to how your friendly neighbourhood Booking Clerk would have done it in the days of printed NFMs:

Look up origin station - Is there a fare given from that station to your destination? If there is, see what the "appropriate" fare is for the journey and sell it.

If not, look for a fare from the relevant "key" station to the destination.

If there's still no joy, look for a key station for the destination, then check for fares from both the origin station and the origin key station to same.

Picking a random example: Chappel and Wakes Colne to the wonderfully named Hall I' Th' Wood, if you look it up, is actually priced from Colchester Stations to Bolton.

Cheers,

Barry

 

[clagmonster]

In terms of a freedom of information request, as the TOCs themselves are responsible for setting fares, I would try making a FOI request to East Coast, as they are currently a public company. I have read somewhere that they have been cooperative with FOI requests so far. The Penzance to Wick route +London fare is set by East Coast.