Croeso 2012

[Jon Brooke]

What would be your solution then Graham?

I like the idea of the ranking system, and as a relative late-comer to orienteering I have been happy to see my progress reflected through it - except that my overall progress (in the ranking system) now seems to have gone into reverse even though I know that I am still improving.

I suppose that my ranking within M45 should at least remain meaningful?

[DeerTick]

As a ranking list anorak, I like these discussions.

My, purely anecdotal, take:

Last year I resigned myself to age taking its toll and the seemingly generally accepted opinion that my age group's scores would drift downwards anyway. I was struggling to score above 1000.

This year, however:

A stinker (too many): 950 ish,
a fair run (a few): 1000 - 1049,
a good one (too few): 1050 - 1099,
an excellent run (none, but could have been): 1100 plus

I'm OK with that.

AP

[graeme]

What would be your solution then Graham?

Well, at the risk of rerepeating myself...
The normal way to start a statistical analysis like this is to guess a starting ranking (as they did), then process the real data/results again and again until the resulting ranking settle down...
1/ Set initial data "ranking list zero"
2/ Use 2011 results to generate new data "ranking list 1"
repeat with same data to generate list 2, 3, 4, 5 until "list N" = "list N+1"

Once that's done, the algorithm in Appendix K should work OK, and all the code should already be written.

Bootstrapping http://en.wikipedia.org/wiki/Bootstrapping_statistics is probably better but more complicated and would require recoding.

[Jon Brooke]

OK - next question(s) (apologies if you've answered these elsewhere too):

Why isn't what you suggest being done?

How come all seemed to be going OK in the ranking system (from my POV at least, noticing a trend that matched my performance, even if my absolute performance was ranked too highly) for a couple of years and it is only now that I am noticing a big correction?

Is BO bovvered?

[graeme]

Why isn't what you suggest being done?

Don't know. I wasn't on the ranking working group, didn't devise the algorithm, and haven't
had a reply from BO when I asked for the code.

My guess is that BO subcontracted the website/code development, didn't fully test it, paid up, and now have no leverage on the contractors to put it right or expertise to do it themselves. I have no evidence for this. but I have a habit of being right

Back in 2007 Mike Cumpstey coded up (my proposal for) a single list and started it in this iterative way. It seemed to work fine, but the ranking working group decided on a slightly different algorithm (which should also be fine if implemented correctly).

I should add that the problems are quite minor compared to some of the more crazy suggestions we see here from time to time (like giving loads of points for bad runs at big events). Remember, the ranking list is trying to answer the question "based on this result how good is Jon Brooke at orienteering": your counting scores are 1180, 1199, 1179, 1175, 1196, 1184 which are remarkably consistent compared with other schemes.

[Jon Brooke]

your counting scores are 1180, 1199, 1179, 1175, 1196, 1184 which are remarkably consistent compared with other schemes.

It's the non-counting scores I'm concerned about.

I did good runs (for me) on 5 out of the 6 days at Croeso that I'm fairly confident were in line with those scoring results, if not better, yet the highest ranking points awarded were 1157 (I think).

[greywolf]

but the ranking working group decided on a slightly different algorithm

No, not the case. THE RWG terms of reference state

The Workgroup is not responsible for:
a. Establishing the details of an operational ranking algorithm
b. The operational procedures that will support the use of the Ranking System

BOF had already decided that the algorithm from the "old" system would be used. Which make their inability to implement it correctly even more pathetic

[Ali Wood]

The Workgroup is not responsible for:
a. Establishing the details of an operational ranking algorithm
b. The operational procedures that will support the use of the Ranking System

...So the Workgroup had to value the second-hand vehicle but weren't allowed to look under the bonnet!!

[Dangerous]

Re BOF ranking points

I noticed a recent change in the BOF rankings:

I moved up approx 150 places whilst my scores were unchanged

The points required to be in the top 25 in my club dropped by almost 300

The points required to be in the BOF top 1000 dropped by 60

It may be coincidence but these changes happened when the Scottish Six Day scores expired.

[DJM]

BOF had already decided that the algorithm from the "old" system would be used.

Not sure this is true! There was a long debate inside the RWP regarding the principles behind the points award and all sorts of options were aired. The previous scheme easily came out on top, in principle that is, but a significant re-work was needed to adapt it to produce the single rankings scale we have today. The Rankings Appendix needed to be completely re-written as a result.

[HarryO]

Re BOF ranking points
I noticed a recent change in the BOF rankings:
I moved up approx 150 places whilst my scores were unchanged
It may be coincidence but these changes happened when the Scottish Six Day scores expired.

That's exactly why you moved up. Me too. There are quite a few people who seem to only enter just the S6D each year - and this will account for a lot of their 'top six' scores. So when the scores drop off a year later, then they dissapear from the top of the ranking table. Especially if they didnt go to Wales.