ranking list

[Scott]

careful! don't forget we abolished second clubs on the membership card/database

As andy suggested, you could (in addition) let people automatically claim results where the name and ecard number match the database, which would catch a few more. People running for a second club with a hired ecard would still be stuck, but it would be an improvement on the current situation.

[madmike]

Scott,
more than content with that.

However, you wrote "name and club" and that would have been fine when both my clubs were on my membership card and the database, which they aren't now - not having a go at you.

I am displeased and dismayed by BOF's "we don't have that functionality anymore" dismissal of the issue and am using your comment to illustrate their oversight!
yours in O
Mike

[EddieH]

Well I entered Margaret and myself via Oentries with our BOF membership numbers being recorded there. I did nothing differently from usual, and nothing differently between the 2 of us. I have points - she does not.

Jason presmably only entered once yet has scored twice out of 3

[Monte]

Also at Edinburgh why do so many seniors not have a score by their name? - I thought perhaps they had not renewed their membership but I notice some cases where that is definitely not the case.

Looking at the entries for Carlisle I note that most are 'Senior BOF Member', some (including me) 'Senior' and some (including Roger and Scotia!) 'Senior Non BOF'. I now notice I was charged the extra £2 for non-BOF. I vaguely remember seeing a box showing "NBM" as the default - not realising what it meant I ignored it.
Even though BOF numbers are in the Oentries system it still requires confirmation that you are a BOF member. Entries for the Scottish Nights have as the default 'Non SOA/BOF'.

I guess that this could cause BOF numbers and Ranking Points to be omitted/lost.

Beware of Defaults.

[Scott]

"NBM"

...which, helpfully, could be "non-BOF member" or "National BOF member" .

[Jon X]

Looking at the Edinburgh City results it demonstrates that it will take a long time (i.e. years) for the short course bias to be ironed out. Because of initial bias the short course points are clearly inflated over the long by quite some margin.

Jason scores 1209 points on the long and would have been well in excess of 1300 on the short had he been given a score for that.

Although as it happens Jason was given the wrong map and ran the short first - he is listed as non-comp on the long as i) he had already run the short and so visited some of the sites and ii) he was knackered! So perhaps not so wrong for his short course score to be higher - more wrong that he doesn't have a score for it

[Spookster]

Well right now Jason is scoring ranking points on all three courses!

Short, 1st place, 1320.
Long, 38th place, 1205.
Yellow, 1st place, 1274.

Perhaps someone needs to have another go at the upload...

(NB there is currently no way to show someone as n/c in the BOF results. N/c runners have to be marked as DSQ by putting a 3 in the Classified column).

[EddieH]

I think my point was missed. It seems that the initial data is SOOO inaccurate that it will take YEARS before it sorts itself out. Just look at Sunday's Concorde Chase - Someone gets more points in long orange than anyone on blue

Shorter courses are currently scoring far higher points. As Graeme points out it needs people running a variety of courses for the anomolies to iron out, but at the same time it is emphasised that one person's anomolous result is unlikely to make much difference.

The problem is that the short course runners, having started this year with inflated points will carry on running shorter courses, thereby continuing the inflated scores for everyone in those classes.

It is only very slowly that the MadMike's of thi world who are prepared to plod around a "deep black" course will increase our true elites to the scores they genuinely deserve.

[Spookster]

Just look at Sunday's Concorde Chase - Someone gets more points in long orange than anyone on blue

Shorter courses are currently scoring far higher points. As Graeme points out it needs people running a variety of courses for the anomolies to iron out, but at the same time it is emphasised that one person's anomolous result is unlikely to make much difference.

The problem is that the short course runners, having started this year with inflated points will carry on running shorter courses, thereby continuing the inflated scores for everyone in those classes.

Since you highlighted the Concorde Chase, I had a look. I'm not a statistician, so please don't tell me I've done it wrong - I just did what seemed to make sense to me!

On each course I looked at the mean points scored by ranked runners, their mean time, and by knowing the course lengths I then plotted their running speeds. See the attached graphic.

It seems that the amount of points that could be scored by someone was independent of which course they chose to run, with the exception of Long Orange and Orange, where the points scored were much too high. But points available on Short Green and Light Green (which are short courses) seemed about right.

Hmmm...

[graeme]

It seems that the initial data is SOOO inaccurate that it will take YEARS

Indeed, but as jac pointed out already several years ago, you don't actually need accurate initial data

Here's what you do...
You take some initial data, whatever.
You take the results for last year and run the algorithm for a year, getting new ranking points.
Then you take those ranking points as initial data, and repeat step 2.
etc. etc.

Then either you get a self-consistent set, independent of the initial data, or the whole thing never stabilizes, which is a strong indication of problems with the algorithm.

[greywolf]

Martin, that's a really useful way of looking at the data. Shows up a couple of things: firstly that it's silly to calculate ranking points for Long Orange and Orange - the Rankings working group never intended that this would be the case - and that secondly, that the system works in broad terms as we would expect, i.e. that more points are scored on the longest/hardest courses where the better orienteers are.

The "problem" is that the the differentiation between the mean points per course is not big enough - the red line should be steeper! So whilst it's not true to say that short courses score more points (if that was the case the top of the rankings list would be dominated by W65S ), it is true that an individual can probably expect to score more points if they run a shorter course.

Does it actually make a huge amount of difference to your position in the rankings? Probably not.

How long will it take for this to disappear? Dunno, but I expect it's happening already - there will be a gradual shuffling of the rankings as described in the other thread.

And will loads of L runners suddenly start running shorter courses to gain extra points? well maybe...but of course the more that do, the quicker the effect will be eroded.

[sketchweppers]

Shows up a couple of things: firstly that it's silly to calculate ranking points for Long Orange and Orange - the Rankings working group never intended that this would be the case - and that secondly, that the system works in broad terms as we would expect, i.e. that more points are scored on the longest/hardest courses where the better orienteers are.

I've just noticed that one of our adult beginners has scored the highest points in our club, by running orange at a couple of recent events. If she carries on at that rate she will soon be at the top of our club's ranking list! Doesn't seem quite right to me.

If the working group never intended TD3 courses to be ranked - why is it happening?

[SYO Member]

EddieH, I may not be typical, but Short course runners won't necessarily "carry on running shorter courses", and Spookster's graph appears to show that it doesn't necessarily matter if they do (excluding Orange). And having just seen Greywolf's posting, it seems that the oppsite may apply and Long runners could switch to Short!

As I have explained on here before, in Badge terms, as a silver standard orienteer with the occasional Gold, when I'm orienteering I like to be out for around an hour, sometimes a bit more, sometimes less. So for many years at district events I have done Blue and at Regionals the Short course.

But having moved up to M55 this year, I find that Long is Blue and Short is Green. So depending on how much info about terrain, course length etc is known in advance before pre-entering at Regionals, I will do a mixture of Long and Short and I know that others will do the same.

And having run M55S at the Northern Champs last weekend (and unfortunately had a poor run), I'm doing M55L (Blue) at NOC's Robin Hood Trophy this weekend.

So excluding the odd individual, it was always a myth that people ran Short just to come high in the relevant Short ranking list (it was just a measure of performance anyway), and we just go orienteering for enjoyment at our level of ability and we're quite happy now to be in a single ranking list for our age category.

Isn't MrsH in a similar situation having also changed age class this year and used to run Short?

[Mrs H]

Yes - I was thinking ditto all the way reading down. I will run green (w55L) at all but multi days. Up until this year this would have been between L and S.

Incidentally - I noticed at the recent Sutton Park event that people who had entered a colour coded event were automatically mapped onto the nearest age class they could have been considered competitive (Hoco chose to run green but was put in the results as running W40S)

This may have been only because it was a West Midlands league event - I hope so - it would have annoyed me to have entered a colour coded and ended up in an age class

[Adventure Racer]

Of course the other problem which has been mentioned, is that even if all the scores are accurate, there will still be a distortion on any course where somebody beats the field by enough to clip the tail of the distribution significantly (under the +100% rule). This could of course happen on a longer course, but is far more likely on a short course. That and the problem with distorted stats with few runners on a course as already discussed means rankings will always be dodgy on low TD courses which only a few ranked runners compete on.

There are solutions to all of these concerns. To make the scores correct do graeme's self consistency thing - you could even go one step further and do that continuously on a rolling basis with the current set of results for the last year, iterating through until you get a stable set of scores (which is what Catching Features does - with an admittedly much smaller ranking list). Lots of computing needed, but surely possible to set up an overnight run. To fix clipping the tail use an algorithm with the mean rather than the best. Simply remove low TD courses from the ranking.

The question being why are we in the situation we're in? None of these things seem like difficult stuff - an option for runners to be ranking n/c, an option for a course not to count for ranking, running some numbers through an algorithm. Did none of this get trialled before release (how difficult would it have been to run the "production system" at the end of last year, get some beta testers to upload results and make unofficial rankings available for Nopesport et al to comment on?)