Results and Rankings.

[DJM]

Jon,

I don't believe that what comes out is garbage. You obviously do, and there we will have to agree to differ.

But "garbage in"?? Could you please defend this ... ?

[distracted]

But "garbage in"?? Could you please defend this ... ?

Simple, really. If the initial seed data is not accurate, everything based on it will be flawed. Given that the seed data was based on individual age classes, with people rarely running across courses, unless you have the "seed formula" (which I remember being based on some mins/k calculation?) bang on there is an inbuilt systematic error due to this initial inaccuracy.

Because of the way the ranking system works, this inaccuracy won't go away easily unless you have a lot of people running across courses, to "average out" the errors in the data.

For example, take a course of 30 runners, 15 of whom are relatively overranked by 20 points each across an average of 9 events, and 15 who are "correctly ranked". This will correct the 'average' score of those overranked downwards by only 5%, and those "correctly ranked" upwards by 5% i.e. there will still be an overranking by 18 points on average.

So in this idealised situation, with the same 30 runners you'd need at least 10 events with a 50% course crossover to remove a 20 point overranking error. And every event where there is no people running across courses slows down this correction. Any error is not going to go away quickly.

However, you now have a year's worth of data created by the new ranking list. There are some instances of people running across courses, making a single list more meaningful. So, as has been suggested before, why not use this data as a new seed - you've 100+ events rather than 10 or so, with people running across courses to provide direct comparison - surely that has to give a more accurate start point?

[andypat]

1177 +69 Andy Paterson CLYDE 1967 M 6065 965, 1004, 1034, 964, 1061, 1037
1177 +11 Andrew Macpherson CLYDE 1967 M 6065 991, 993, 987, 1033, 997, 1064


Anybody know what criteria is used to differentiate between competitors on the same points? It cant be alphabetical.

For some reason my suggestion that its based on looks isnt going down to well with my colleague 180 minute man .......

[awk]

Jon,

I don't believe that what comes out is garbage. You obviously do, and there we will have to agree to differ.

But "garbage in"?? Could you please defend this ... ?

That's not how I read it. Jon simply quoted the standard phrase meaning that whatever quality of data you put in dictates the quality of the output. So if the data you put in is only X% accurate, then the output can only be X% accurate at best. I don't believe the output is garbage, but it is certainly flawed in some areas, so the output is inevitably flawed, and needs more correction than the latest changes managed.

[madmike]

Andypat,

I don't know but try BOF number comparison as there are no obvious patterns for ties on the list, including me btw, which eliminates looks, lighter and younger as the deciding factor

[AlanB]

1177 +69 Andy Paterson CLYDE 1967 M 6065 965, 1004, 1034, 964, 1061, 1037
1177 +11 Andrew Macpherson CLYDE 1967 M 6065 991, 993, 987, 1033, 997, 1064


Anybody know what criteria is used to differentiate between competitors on the same points? It cant be alphabetical.

For some reason my suggestion that its based on looks isnt going down to well with my colleague 180 minute man .......

Next best (ie 7th) ranking score?

[andypat]

Could be but seems unlikely as my scores are more variable and I have a couple of really bad ones!

Anyhow - its a moot point for me since I have the Edinburgh race counting twice at the moment. Apart from the sly dig at my compatriot, I was thinking this could actually really matter if it occurred nearer the top - I wonder if there isnt scope for having equal rankings or is that too complicated.

[Roger]

1177 +69 Andy Paterson CLYDE 1967 M 6065 965, 1004, 1034, 964, 1061, 1037
1177 +11 Andrew Macpherson CLYDE 1967 M 6065 991, 993, 987, 1033, 997, 1064
...
I wonder if there isnt scope for having equal rankings or is that too complicated.

Those ranking positions look pretty equal to me!

I suspect that if the points are equal no further tie-break is applied and the listing order is just what comes out of the database query.

[andypat]

Good point Roger - in my haste I had assumed that as I was on top I was ranked above him!

[schnitzer]

Maybe what I am about to say is an echo, if so I apologise but I'm not about to read 30 pages of posts.

I am wondering what the goal is of the continued work on the ranking system. Looking at just the first 100 I can't really work it out. It doesn't really look like a handicap system but then it doesn't really look realistic either. I can understand the fascination (and I admit I would quite like to have a go at different mathematical approaches with the data myself) but surely in the end it is just guess work i.e. nothing more than a bit of fun? Apparently not according to Mike Hamilton..

We believe that the ambition to award rankings points to all eligible runners on the same single scale, irrespective of (eligible) course completed, has been more than satisfactorily achieved and that British Orienteering now runs the most comprehensive and sophisticated rankings system anywhere in the world

I can't help feeling that BO has sold its soul to the advertising/marketing devil, but then I've said all that before.

[andypat]

No - that would be the Coca Cola British Ranking list.

But to be honest I cant think why the ranking list couldnt be sponsored by some O related business.

Personally I think the ranking list is much better than previously and people on here seem to have a pretty good handle on what could be done to improve it. At the top end I am sure that everyone knows thats not necessarily the order they would finish in an event, but its still great value and motivation for the rest of us also rans.

[AlanB]

Maybe what I am about to say is an echo, if so I apologise but I'm not about to read 30 pages of posts.

I am wondering what the goal is of the continued work on the ranking system. Looking at just the first 100 I can't really work it out. It doesn't really look like a handicap system but then it doesn't really look realistic either. I can understand the fascination (and I admit I would quite like to have a go at different mathematical approaches with the data myself) but surely in the end it is just guess work i.e. nothing more than a bit of fun? Apparently not according to Mike Hamilton..

We believe that the ambition to award rankings points to all eligible runners on the same single scale, irrespective of (eligible) course completed, has been more than satisfactorily achieved and that British Orienteering now runs the most comprehensive and sophisticated rankings system anywhere in the world

I can't help feeling that BO has sold its soul to the advertising/marketing devil, but then I've said all that before.

I think the goal of the continuing work on the ranking system is to further refine and improve it. Since being launched people have identified a number of anomalies and discrepancies, and the group responsible for the system have sought to identify the causes and take action.

The ranking system isn't meant to be a handicap system, although it would be interesting to see any suggestions as to how it could be the basis for one! As I understand it the BOF Rankings seek to list British orienteers by the quality of their performances over a twelve month period. It measures the quality of each performance by comparing it with the mean finishing times of competitors, the spread (standard deviation) of the finishing times and the standard of the participants in the race - as measured by previously won ranking points.

Therefore, in theory, I can measure this week's performance on a green course with last weeks M50L championship run the week before. The green course was a fairly low-key affair with mainly local runners, the championship race contained some of the best competitors at my age class, yet looking at the calculated points I can see that my performances were broadly equivalent in terms of quality. I can't tell that from looking at the finishing times (vastly different distances) or by looking at my finishing position for each week.

It is a handy tool for observing improvement over time and enables comparison to be made with others who run different classes and courses. It is not perfect, and some anomalies will always happen, but I find it an extremely useful guide as to how I am doing; and, considering the fuss made when a delay in posting the results happens, so do many others! The existence of the rankings has certainly made me compete in ranking standard events more often and, get this, encouraged me to train and work on my navigational abilities - good things, yes?

For those of us who are never going to come near winning anything, seeing yourself improve your ranking position against 'rivals' you may never race against is very motivational.

As for BOF selling their soul (our souls?) to the 'advertising/marketing devil', a) I'm not sure what you mean, and b) what the relevance is to a ranking system based upon statistically comparing performances.

[graeme]

British Orienteering now runs the most comprehensive and sophisticated rankings system anywhere in the world

Can't disagree with Mike there.
Since he didn't use the word accurate

I can't help feeling that BO has sold its soul to the advertising/marketing devil, but then I've said all that before.

O tempora, O mores:
It really would be appalling the ranking list were sponsored by, say, a wine merchant.
Like it was in the 80s

[schnitzer]

1. I'm not having a go at what the ranking list provides. I was one of those people really annoyed when my scores don't go up quickly when I've had a good run, using it to gauge my improvement in orienteering.

2. I also agree that the single list was a good idea to support the regional event age classes to colour scheme idea i.e. run what you want... and now still get ranked.

3. The continued work - fantastic! I would love to be one of those involved in trying to draw the best statistics from the raw data. Sounds like a challenge!

So I basically agree with all of your post AlanB so then I just need to clear up what I meant by 'advertising/marketing devil'. My point was that when I read Mike's piece on the BO website the way it was worded made me expect to see a masterpiece of perfection when I loaded up the rankings page when in fact the basic problems (see 'garbage in, garbage out' - a bit harsh but basically fair) that existed before the change to a single list have been expanded by that change and it looks (IMO) a combination of both handicapped and realistic.

I think Graeme's 'accurate' point sums it up perfectly. Mike has worded this like a salesman or a politician. 'It's propaganda, sell it to the innocent, they're buying it' but at the same time make sure it's not telling any lies. Yes Graeme, maybe things have been this way for a long time, maybe things are better than they used to be but it just makes me disappointed that in this world of hype and fake that even my beloved hobby of orienteering isn't immune from the virus that is advertising/marketing. Why can't Mike have just simply said 'we've had a good go and will continue to try to improve it but obviously it's not perfect - it's definitely been interesting and a lot of fun!'.

[Big Jon]

Graeme

you're learning cynicism - must be catching it from those nasty Nopesport posters.