One Tonner winner results changed for seeding

[DMC007]

Please, please please let it be a mistake when they did the entry

 

..relax SP, you will get to where you want to be in due course. Dont climb that ladder too fast!!

[nuge]

well done PPA - finally adjusted the argus seeding index to make a little more sense.

 

i'll stop moaning now! cheers...

[headhunter]

well done PPA - finally adjusted the argus seeding index to make a little more sense.

 

i'll stop moaning now! cheers...

 

Yeah, I noticed that as well..... there was in my case; a 7 point swing in my favour.

 

WTF, this surely proves that this PPA seeding system is indeed f@rked up ?

 

And while I'm bitching; there was a 5 point difference between my PPA index and CL national index before the 1 Tonner and yes you guessed it..... they both used the same 3 races to provide me with an index....now that truly is f@rked up.

Edited October 28, 2010 by headhunter

[niklasvm]

The algorithm is still weird though - I got better indeces for random races I rode last year where I started in M and was pretty unfit than for the Westcoast and Engen races where I came top 10 in my bunch (H). For example, in the Westcoast I got a better time than anybody in the preceding bunch, came 4th in my group (2s off the winner) and still only got an H equivalent index. This means that for some races its effectively impossible to improve your seeding.

 

Fortunately I did decent One Tonner and can move up to F

same things happened to me but as soon as you get to start riding in groups F-H you get stuck as those groups can never close the gap between the group time and the winners time. i used to ride in group G as we consistently did better times than F. in the simonsvlei race we even overtook the F and E bunches and finished the race with the D's. This still only gave me a seeding index equivalent to an F.

 

the best advice to beat the seeding system is train hard for the summer season and make sure you are light and fast on the climbs, then you can ride away from your group and the other wheelsuckers and hopefully gain some time on the winner.

 

 

[stratus5]

I wonder if rather than all the linear regression data fits against the Argus time etc etc, thy don't just take the winner of the G group and assign a delta such that the winning time stays just inside G.

 

Last season G/H looked to be incredibly strong. They even caught D a few times. Don't think many of them were promoted because G/H looks even stronger this year.

['Dale]

I'm confused like some of you.

 

My top 3 races are still the same and is used to determine my seeding.

And I got bumped up from J to I...

 

Huh?

[Tankman]

Yeah, I noticed that as well..... there was in my case; a 7 point swing in my favour.

 

WTF, this surely proves that this PPA seeding system is indeed f@rked up ?

 

And while I'm bitching; there was a 5 point difference between my PPA index and CL national index before the 1 Tonner and yes you guessed it..... they both used the same 3 races to provide me with an index....now that truly is f@rked up.

 

How f@rked up is this? Check my results of the Westcoast for this year and last?!

 

 

Last year the race was 5km shorter and only 1 second difference in the winners adjusted time for both races?

Difference in my time is only 43 seconds but the index is way different?

I would have thought this years longer race should give a better index?

 

This is why we need a National Seeding System - NO More PPA!

[Woofie]

How f@rked up is this? Check my results of the Westcoast for this year and last?!

 

 

Last year the race was 5km shorter and only 1 second difference in the winners adjusted time for both races?

Difference in my time is only 43 seconds but the index is way different?

I would have thought this years longer race should give a better index?

 

This is why we need a National Seeding System - NO More PPA!

 

[headhunter]

Skankman and Woosie.....there's a movie in there somewhere? A 1970's blue move ....que the piped organ music :D

Edited October 29, 2010 by headhunter

[niklasvm]

I wonder if rather than all the linear regression data fits against the Argus time etc etc, thy don't just take the winner of the G group and assign a delta such that the winning time stays just inside G.

 

Last season G/H looked to be incredibly strong. They even caught D a few times. Don't think many of them were promoted because G/H looks even stronger this year.

 

here's an idea: think about it, your seeding decides which group you are riding in. if you are faster than your group, you should be moved up. the onus though is on the stronger guys in a group to ensure they drop the rest of the wheelsuckers. they could maybe adjust the seeding for each group so that the average time falls in the middle of the seeding index. then if your "time" falls ahead of the expected group then you should be moved up.

 

the biggest problem with the model is the fact that it assumes our race times to be normally distributed when they are clearly skewed

[Bram_on_Madone]

here's an idea: think about it, your seeding decides which group you are riding in. if you are faster than your group, you should be moved up. the onus though is on the stronger guys in a group to ensure they drop the rest of the wheelsuckers. they could maybe adjust the seeding for each group so that the average time falls in the middle of the seeding index. then if your "time" falls ahead of the expected group then you should be moved up.

 

the biggest problem with the model is the fact that it assumes our race times to be normally distributed when they are clearly skewed

 

Agreed. The linear regression model they use, ignores your starting group and is based on normal distribution of time across the starting groups. It is also adjusted for the relative of spread of everybody that did the Argus - this is in truth a "randomness" factor, as it all depends on how many people that rode the last Argus are in any particular race and how they happen to be spread.

 

Last year I did the Burger and the Engen in almost the identical time and the winning time was very close to each other, but I got a much better seeding for the Engen, just because of this randomness factor.

[Woofie]

Skankman and Woosie.....there's a movie in there somewhere? A 1970's blue move ....que the piped organ music :D

 

I was hearing the lame techno music there instead

Edited October 29, 2010 by Woofie

[Woofie]

Skankman and Woosie.....there's a movie in there somewhere? A 1970's blue move ....que the piped organ music :D

 

My wife came up with a pretty good one as well....

 

Spankman

 

Goes pretty well with the current theme.

[niklasvm]

Agreed. The linear regression model they use, ignores your starting group and is based on normal distribution of time across the starting groups. It is also adjusted for the relative of spread of everybody that did the Argus - this is in truth a "randomness" factor, as it all depends on how many people that rode the last Argus are in any particular race and how they happen to be spread.

 

Last year I did the Burger and the Engen in almost the identical time and the winning time was very close to each other, but I got a much better seeding for the Engen, just because of this randomness factor.

exactly and the variance of the league race times is heavily biased towards the racing category riders meaning an even more distorted seeding calculation.

[PeterM]

I am just glad that I am not the only one who finds the way the seeding is done, both irritating and frustrating.

 

I can see two key problems:

1. There seems to be no consistency – race results are adjusted, to the extend, that our seeding results are not sufficiently comparable, nor fair.

2. There is also a lack of transparency – what we are doing right now is just making “educated” guesses. If it would be clear how a race is adjusted than we “would know”. Why are the individual race seeding/adjustments details a secret? They should be on the website for all to see. Those non-disclosures just cause suspicion and anger.

 

Maybe its time to have a nationally consistent system. And for practical reasons it should only be changed/adjusted on yearly or half-yearly base.

[Woofie]

here's an idea: think about it, your seeding decides which group you are riding in. if you are faster than your group, you should be moved up. the onus though is on the stronger guys in a group to ensure they drop the rest of the wheelsuckers. they could maybe adjust the seeding for each group so that the average time falls in the middle of the seeding index. then if your "time" falls ahead of the expected group then you should be moved up.

 

the biggest problem with the model is the fact that it assumes our race times to be normally distributed when they are clearly skewed

 

It used to somehow work a bit better last year.

 

Last year, if you finished high up in your group you would generally get a better seeding.

Middle of your group you would stay there

And end of your group a bad seeding (hense drop)

 

So going into a race knowing that if you stick to it to the sprint it would normally make you good enough to up your seeding, but now you are going to stay exactly where you are as the winners of the group are not cutting it.

 

Eventually yes, we are all going to be in the same boat and have bad seedings.

I dont really care if there is a A, G or Z on my back, but it is going to be really sad if I wont be able to enter league next year if my seeding drops a few groups from these funny calculations.

 

I am in the same boat as you Tank, I rode the WCE 5 km/h faster this year but I still ended up with pretty much the same ride index as last year.