Rankings query

[Tim]

I suspect this has been done to death but please bear with me as I’m a (sort of) newbie.

Having recently re-joined BOF after a long gap, I am now the proud holder of 1155 ranking points from Sunday's exertions. Having an idle moment and being mildly interested, I bothered to look at how they are earned. One thing confuses me – it appears the case that you gain points no matter your age class and the course you run. So if, as an M40, I slaughter a load of well ranked M10s on a white course I get loads of points?

Doesn’t seem right or have I misunderstood.

[Scott]

In simple terms, the ranking scores are determined both by how you run relative to everyone else on the course and by the previous ranking scores of everyone on the course. In theory, the same "quality of run" should deliver the same number of ranking points regardless of which course you're on and who you're racing against.

Whether the system actually works quite like that in practice has been the subject of prolonged discussion on here, but although there are some oddities, and the occasional freak score, it seems to do a reasonable job most of the time.

[smf]

So if, as an M40, I slaughter a load of well ranked M10s on a white course I get loads of points?

Doesn’t seem right or have I misunderstood.

No only top year M/W16 and over are ranked.

and

Courses where there are fewer than 10 ranked runners are not included in any rankings calculations.

http://www.britishorienteering.org.uk/i ... ndix_k.pdf

[Tim]

No only top year M/W16 and over are ranked

Replace M10 with W60 then.

In the olden days you only gained ranking points at badge events, national events and championships and therefore were only ever compared against people of your own age class. It just appears odd that at the Chiltern Challenge, had I been a member of BOF, I'd have got over 1000 points for running a Green course, when as an M40 I ought to have been doing Brown.

[greywolf]

You've misunderstood.
Firstly, under 16s aren't including in the rankings. (many of us think they should be if they are doing TD 4 or TD 5 courses but that's a different issue)
Secondly, in theory, it shouldn't matter who you run against: the system allocates points based on an algorithm that compares your performance on the day vs your competitors against the distribution of previous ranking scores - if you run against (and are beaten by) 10 good M21s then because their previous ranking scores are high, you should score as many points coming last as if you were running exactly the same course in the same time against 10 (relatively lowly ranked) W55s and winning by miles. Obviously if you beat the 10 highly ranked M21s you'll score a shedload of points.

Unfortunately in practice, because of the way the system was seeded, if you share your course with W55s you'll probably get more points than if you share it with M21s. This effect may or may not disappear over time, it's pretty difficult to tell as it appears that BOF haven't bothered to implement the algorithm correctly and don't seem to be all that arsed about sorting it out

[King Penguin]

Also, not all events - only "normal" level A, B & C.
No ranking points from any Level D events, or Score / mass start / chasing start / night events.

[greywolf]

In the olden days you only gained ranking points at badge events, national events and championships and therefore were only ever compared against people of your own age class.

Actually no - you were ranked in whatever was the default age category / class of the race in which you competed - at one point I was ranked in 5 different classes, including two which I had never entered

The main change is that under the old system, where a single course was used by multiple age classes, the rankings list would pretend that there were multiple races going on - because these "separate" races inevitably had tiny fields this led to ridiculous anomalies - the new system considers the course as a single race

[graeme]

To get the same number of points, you need to finish mid-pack among M21s or miles ahead of the W60s. The ranking list attempts to give equivalent points for equivalent runs.
So if you run a 5km course in 40 mins it shouldn't matter if the other competitors are labelled as "M21E", middle" and doing 35mins or "W60L" and taking an hour: you should get the same points.

Obviously no system is perfect, but compared with things like "position on course" or "mins per km" the ranking list is way superior. For me, decent runs typically get the same score to within 2-3%.
Even though the system isn't ideal and isn't implemented properly.


Edit : "What greywolf said."

[Tim]

Greywolf - if you ran out of class you should have been n/c unless you were running up which could mean you could be ranked in multiple classes. But the point was you were compared with the people in the age class you entered. The fact that some classes shared courses used to be coincidental.

I understand the system works well when you look at one course in isolation and if people generally run courses corresponding to what you would expect their age-class to run but events have more than one course and you are free to run any course you like.

Sticking with the Chiltern Challenge as an example - (for whatever reason) I ran Green competing mainly against M/W 55-70s, so my ranking points would reflect their performance and current ranking points with no adjustment for the fact I'm not 55-70 but 40 and therefore should be faster than them. If rankings are a reflection of relative performance rather than absolute performance then my points should be downgraded. Again in the olden days (and apologies for harping back) if I remember right, points from S/B courses had a 0.8 factor applied because it was felt running a shorter distance was easier/less worthy. So I walk away with 1000 points for 43 minutes work. Had I done Brown I'd have needed to do the same speed over 90 minutes to get the same points, which I'd suggest is harder.

[Scott]

I ran Green competing mainly against M/W 55-70s, so my ranking points would reflect their performance and current ranking points with no adjustment for the fact I'm not 55-70 but 40 and therefore should be faster than them.

The "adjustment" comes from the fact the points you score reflect their current ranking points. The average M/W55-70 will have lower points than the average M40, and so the points you score will be lower than if you had beaten a load of other M40s by the same margin.

[Tim]

Sorry, I have misunderstood. I assumed the system would be designed to allow a really really good person in any class to have the same number of points as a really really good M21, as the system would try to be a measure of relative rather than absolute performance. But if an average M40 has more points than an average M/W55-70, the system is weighed (however slightly) towards absolute performance.

[DJM]

it's pretty difficult to tell as it appears that BOF haven't bothered to implement the algorithm correctly and don't seem to be all that arsed about sorting it out

Just for the record, I'm currently working with Mike Hamilton to sort this out, with advice from graeme and David Rosen.

Watch this space!

[Jon Brooke]

For me, decent runs typically get the same score to within 2-3%.

Except at the Welsh 6 days

[King Penguin]

Sorry, I have misunderstood. I assumed the system would be designed to allow a really really good person in any class to have the same number of points as a really really good M21, as the system would try to be a measure of relative rather than absolute performance. But if an average M40 has more points than an average M/W55-70, the system is weighed (however slightly) towards absolute performance.

Your original assumption was incorrect - it is more of an "absolute" than "relative" measure.
As a consequence :
top W21 is 64th overall
top M60 is 204th overall

For an estimation of age-adjusted "rankings" see Dave Nevell's article in an edition of CompassSport earlier this year.

[mharky]

"estimation" is an interesting way of putting it