Results and Rankings.

[ifititches]

W40 eh? Mrs H.

By that equation I'd now be W35 after Dymock. Wouldn't have minded being transposed across into that age class, even if I would still have been last. Ah the re-juvenating aspects of the sport, we should sell it as the elixir of life with bramble scratches as an additional previously unknown homeopathic remedy for almost anything.

Lumpy Lycra could even have had a free sex-change from her course, as well as re-juvenation

But re your comment, I just assumed it was the system being a bit clever and matching up age with course run, the same as it must have done for Babs Ford

[EddieH]

There are now enough events around to see that the seed data bias towards shorter courses is very significant. There is a massive advantage in running a shorter course. Jst look at the leading green competitor that you know on any course and then see who is scoring the same on black or brown. I think you'll see that their points are well over-inflated. I am sure that time will change this but it will take well over a year because today's events are perpetuating the bias.

Incidentally there are still some really weird anomolies. An M14 has been allocated 1026 points on Orange at Gullane

[andy]

interesting to see that an event which will have one of the best elite attendances of the year is a 'local' event which won't be ranked unless changed:

BOF Event Page
Start List

[frog]

Woops went to it and saw the closing date for it and the SOL was yesterday. I hate events with closing dates more than a week in advance. Will email the organiser.

[Jon X]

interesting to see that an event which will have one of the best elite attendances of the year is a 'local' event which won't be ranked unless changed:

BOF Event Page
Start List

Don't worry Andy we have spotted that too - we are looking to upgrade it to Level 2 specifically because it feels like it would be inappropriate for it not to be included in the ranking system.
(it is no longer quite the 'little Saturday middle race' we envisaged when it was originally registered!)

[Timsands]

I recently returned from the Iberian peninsular having enjoyed 4 weekends of extremely well organised orienteering in excellent and varied terrain where nothing is too much trouble. Thankfully my time away has saved me from exposure to the all new single ranking list until now. To my surprise I seem to be doing rather well. That is if I stick with the Blue course.Take last week I had an absolute disaster at the Wimborne event yet scored 1128 points. Had I run the same minutes per km on the Brown I calculate I would be lighter by 200 or so points. This weekend I did even better, over 1200 points. To get an equivalent score on the Brown course I would need to run 1.5min/km faster. Applying my min/km logic to the Green course, perhaps I could have scored even higher. Not being a Maths Master, perhaps some kind person can explain why.

[Spookster]

Take last week I had an absolute disaster at the Wimborne event yet scored 1128 points. Had I run the same minutes per km on the Brown I calculate I would be lighter by 200 or so points. This weekend I did even better, over 1200 points. To get an equivalent score on the Brown course I would need to run 1.5min/km faster. Applying my min/km logic to the Green course, perhaps I could have scored even higher. Not being a Maths Master, perhaps some kind person can explain why.

I'm not a maths master either, but I think the issue is that the system is still "settling down". There are some issues around the correctness of points scored by runners on courses that have very few ranked runners, but that isn't what you're noticing.

Attached graph shows minutes/km versus points scored, for all courses at the Wimbourne event. Ideally, all the lines should be parallel and on top of each other, which means that you score the same number of points by running at a certain minutes per kimoetre, whichever course you are on.

WIM-points.JPG (70.24 KiB) Viewed 5040 times

You are right, you would have scored fewer points by running 10 mins/km on Brown (perhaps only 950 points) or Short Brown (about 1070 points). But you wouldn't have scored more points at 10 mins/km on Green or Short Green - on those you would still have got about 1130.

Browns and Blacks do seem to quite often get "under-scored", though not at all events. As has been suggested previously, we need a few events where lots of people are running the same course, in order to try to get everyones' relative scores "settled" better. Give it time.

[Simon]

Ideally, all the lines should be parallel and on top of each other, which means that you score the same number of points by running at a certain minutes per kimoetre, whichever course you are on.

Really? I'd expect to have to run a faster speed on a shorter course to get the same number of points, due to being able to sustain a faster pace over a shorter course.

[Spookster]

Really? I'd expect to have to run a faster speed on a shorter course to get the same number of points, due to being able to sustain a faster pace over a shorter course.

That may turn out to be the case, but it would be based on the average time & average points of the people running the shorter course, not because the points formula in the system requires that. The formula does not take account of course distance.

What I meant by "ideally all aligned" is that it one course shouldn't be way off in comparison to the others, as has been the case at a few events.

[DJM]

Comparing mins/km across different courses at the same event can be a hugely inexact science, and may provide much, if not the whole explanation for Tim's apparent discrepancies.

The first problem is that it is far better to compare mins/km rates when distances have been corrected for height climb (by adding 0.1 km for each 10m of climb). However, height climbs are not published for Cranes Moor so we can't do this.

Secondly, different courses are rarely of the same average physical difficulty, they have different average leg lengths, and the ratio of actual distance run to the straight line distance published will not be the same either. All of these will affect running speeds.

So it's difficult to draw firm conclusions by comparing mins/km across different courses at just one event ...

[distracted]

So it's difficult to draw firm conclusions by comparing mins/km across different courses at just one event ...

True, but isn't mins/km the way you 'seeded' the list in the first place? The trend is the same across many of the events, and with the current system won't dramatically improve. As the ranking list works based on average scores, you'd need a lot of people running across different courses to act as comparators and change the mean points of runners to what they should be.

For example, someone who is averaging 1000 points across 10 runs on their 'normal course', then only scores 950 in the next because they run against a different group of people on a different course combination, is only going to have their average reduced to 995...

Letting the list 'settle down' is not going to solve the problem - the algorithm used isn't convergent. If you want it to be as accurate as possible, the list needs a few more months of results to get enough data with people running across courses, and then a convoluted calculation to get everyone's average score "correct" relative to everyone else's...

[greywolf]

Take last week I had an absolute disaster at the Wimborne event yet scored 1128 points. Had I run the same minutes per km on the Brown I calculate I would be lighter by 200 or so points. This weekend I did even better, over 1200 points. To get an equivalent score on the Brown course I would need to run 1.5min/km faster. Applying my min/km logic to the Green course, perhaps I could have scored even higher. Not being a Maths Master, perhaps some kind person can explain why.

As Spookster has demonstrated, your sums for the Wimborne event are slight overestimates but in the right ballpark, however this week:
Blue 7.725 km +75m climb in 52.01 is 6.14 mins/corrected km for 1211 points
Same speed on the 9.7km +80m climb brown course would have been 64.26, which would have scored 1198 points...so there is a difference, but nowhere near the scale you're suggesting.

The problem arises because the original list was seeded according using mins/corrected km, without taking any longer=slower effects into account (or that courses for old folk might stick to physically easier terrain), therefore the ranking list overrates those who usually run shorter courses (note: this is not exactly the same as those who run "S" courses) i.e. oldies, women etc. Your ranking points from any event reflect the competition - if you run against those who are overrated then you will score a higher score for the same performance than if you run against those who are underrated (most obviously, M21).

[AAH]

I think the effect is mostly due to the factor Greywolf has described… but I think there is also a small course effect, or at least, an effect due to the emphasis on outliers.

The runner who came 13th on Brown had a ranking just short of average for the field an average of his 6 best score of 1161 vs 1183. This will be a bit wrong as we don’t have access to the average scores of runners – just their quoted points on the current list - but on any calculation it seems likely he was expected to be just short of average for the field and might have been expected to take 80:30 rather than the 97:57 which he did take. So his time will have had a substantial effect on the standard deviation of runner’s time (from about 10.8 to about 12.7 mins).

Without access to the average score of runners we can’t calculate this accurately but using actual ranking list scores, and mixing in 30% of the small course formula which I believe you are meant to do for 11-19 runners, my best guess is that the slower than expected run by the 13th place runner affected the score of the 11th place runner (very similar mins/k to Tim) by about 25-30 points. (I think the effect on the score of the first placed runner was minimal though).

This is not a huge effect, but is significant…

[frog]

How do you find the list of events that have been ranked? It used to be easy on the old list but I can't find them on the current one. It also says if I hover over a score it will show me which event the score refers to but nothing happens on this computer (old work one)

[Spookster]

How do you find the list of events that have been ranked? It used to be easy on the old list but I can't find them on the current one. It also says if I hover over a score it will show me which event the score refers to but nothing happens on this computer (old work one)

On the BOF website, click Event Information, then Results.

http://www.britishorienteering.org.uk/event/results.php

That lists all the events that have had their results uploaded. If they are Level 2 or 1 (and not the excluded events like night or mass-start) then they will be ranking events. Click the Results links to see the points scored.

The mouseover pop-up works in IE7. Perhaps you have something older?