Compare rankings

From the CRASH-B's to an online challenge, discuss the competitive side of erging here.
User avatar
sjors
2k Poster
Posts: 300
Joined: July 3rd, 2016, 3:50 am
Location: Netherlands
Contact:

Re: Compare rankings

Post by sjors » September 11th, 2020, 12:18 pm

Just corrected a bug, where the Slides and the Dynamic rower got swapped.
Siebe Jongebloed
57/M/1,92/98 kg (getting back into shape again)
Image

Smart rankings: https://www.rankedworkouts.com/

User avatar
sjors
2k Poster
Posts: 300
Joined: July 3rd, 2016, 3:50 am
Location: Netherlands
Contact:

Re: Compare rankings

Post by sjors » September 11th, 2020, 5:13 pm

Thanks to Bjørn Johansen who pointed me out to https://de-maas.nl/correctiefactoren/be ... toren.html (In dutch language only), I think I have found the formulas I need to calculate a age-independend-score.

To clarify
They divide 2 groups of performances: short= < 4km and long > 4 km

Symbols used in formulas:
y= factor to calc max performance
x=age

For Men:
age: 13 -22, short and long:
y = 3,99843E-07*x^6 - 3,89632E-05*x^5 + 0,001508046*x^4 - 0,028884498*x^3 + 0,270552534*x^2 - 0,917594593*x

age: 23 - 26 short and long:
y = 1,000 *

age 27-91 short:
y = 4,67930E-12*x^6 - 1,58674E-09*x^5 + 1,80622E-07*x^4 - 9,14436E-06*x^3 + 1,62205E-04*x^2 - 5,95983E-04*x + 1

age 27-91 long:
y = 3,70323E-13*x^6 - 3,00914E-10*x^5 + 3,20403E-08*x^4 - 1,03748E-06*x^3 - 3,50846E-05*x^2 + 1,11544E-03*x + 1

For Women:
y = 0,9 * factor for men.
Siebe Jongebloed
57/M/1,92/98 kg (getting back into shape again)
Image

Smart rankings: https://www.rankedworkouts.com/

User avatar
MudSweatAndYears
Paddler
Posts: 27
Joined: May 24th, 2020, 6:31 am

Re: Compare rankings

Post by MudSweatAndYears » September 12th, 2020, 12:32 pm

sjors wrote:
September 11th, 2020, 4:35 am
Thanks Dino.
I have taken noticed of how Nonathlon works. I assume they calculate with watts as well. And maybe also therefore as faster you get the more each second increases your points???
The problem I am/was? trying to overcome is the fact that when there are relatively less performances in a specific class the higher your score tends to be. It is seldom you see a Heavy weight younger than 60 in the top 10 for years. (Probably some little personal frustration is coming up here. :D)
The Nonathlon seems to be skewed in favor of females: over the 19 seasons completed, 16 times a female won. So it should not come as a surprise that a specific male category (heavy weight younger than 60) is under-represented.

Also, the scoring mechanics in the Nonathlon is quite complex and therefore less transparant.

Despite its flaws, over a period spanning almost two decades, the Nonathlon continuous to be successful in stimulating C2 users to compete. But a better (fairer and more transparant) handicap system (like you are seeking) would be a big step forward.
I run in the mud, I sweat on the erg, and I happily battle the years...
M 59, 1.83m/6', 74kg/163lb. Aiming for sub-40 on 10k row and on 5k+5k row&run duathlon.

User avatar
MudSweatAndYears
Paddler
Posts: 27
Joined: May 24th, 2020, 6:31 am

Re: Compare rankings

Post by MudSweatAndYears » September 12th, 2020, 12:41 pm

sjors wrote:
September 11th, 2020, 5:13 pm
Thanks to Bjørn Johansen who pointed me out to https://de-maas.nl/correctiefactoren/be ... toren.html (In dutch language only), I think I have found the formulas I need to calculate a age-independend-score.

To clarify
They divide 2 groups of performances: short= < 4km and long > 4 km

Symbols used in formulas:
y= factor to calc max performance
x=age

For Men:
age: 13 -22, short and long:
y = 3,99843E-07*x^6 - 3,89632E-05*x^5 + 0,001508046*x^4 - 0,028884498*x^3 + 0,270552534*x^2 - 0,917594593*x

age: 23 - 26 short and long:
y = 1,000 *

age 27-91 short:
y = 4,67930E-12*x^6 - 1,58674E-09*x^5 + 1,80622E-07*x^4 - 9,14436E-06*x^3 + 1,62205E-04*x^2 - 5,95983E-04*x + 1

age 27-91 long:
y = 3,70323E-13*x^6 - 3,00914E-10*x^5 + 3,20403E-08*x^4 - 1,03748E-06*x^3 - 3,50846E-05*x^2 + 1,11544E-03*x + 1

For Women:
y = 0,9 * factor for men.
I put these age-handicap correction factors in a spreadsheet and plotted the correction factors as function of age. These I compared with the plot available at the Nonathlon web page. The 'De Maas' correction factors are significantly more conservative compared to the corrections applied in the Nonathlon competition (higher ages receive less compensation). Not sure if the same applies to the correction factor for gender.
What is missing from the 'De Maas' tables and formulas is a rowers weight correction factor.
I run in the mud, I sweat on the erg, and I happily battle the years...
M 59, 1.83m/6', 74kg/163lb. Aiming for sub-40 on 10k row and on 5k+5k row&run duathlon.

User avatar
hjs
Marathon Poster
Posts: 9308
Joined: March 16th, 2006, 3:18 pm
Location: Amstelveen the netherlands

Re: Compare rankings

Post by hjs » September 13th, 2020, 4:51 am

MudSweatAndYears wrote:
September 12th, 2020, 12:41 pm
sjors wrote:
September 11th, 2020, 5:13 pm
Thanks to Bjørn Johansen who pointed me out to https://de-maas.nl/correctiefactoren/be ... toren.html (In dutch language only), I think I have found the formulas I need to calculate a age-independend-score.

To clarify
They divide 2 groups of performances: short= < 4km and long > 4 km

Symbols used in formulas:
y= factor to calc max performance
x=age

For Men:
age: 13 -22, short and long:
y = 3,99843E-07*x^6 - 3,89632E-05*x^5 + 0,001508046*x^4 - 0,028884498*x^3 + 0,270552534*x^2 - 0,917594593*x

age: 23 - 26 short and long:
y = 1,000 *

age 27-91 short:
y = 4,67930E-12*x^6 - 1,58674E-09*x^5 + 1,80622E-07*x^4 - 9,14436E-06*x^3 + 1,62205E-04*x^2 - 5,95983E-04*x + 1

age 27-91 long:
y = 3,70323E-13*x^6 - 3,00914E-10*x^5 + 3,20403E-08*x^4 - 1,03748E-06*x^3 - 3,50846E-05*x^2 + 1,11544E-03*x + 1

For Women:
y = 0,9 * factor for men.
I put these age-handicap correction factors in a spreadsheet and plotted the correction factors as function of age. These I compared with the plot available at the Nonathlon web page. The 'De Maas' correction factors are significantly more conservative compared to the corrections applied in the Nonathlon competition (higher ages receive less compensation). Not sure if the same applies to the correction factor for gender.
What is missing from the 'De Maas' tables and formulas is a rowers weight correction factor.
Nonathlon is very much in favor for the oldies, which can’t be right. The olympic skiff champ could not win this competition. Its not for nothing that almost no younger people use this tool.

User avatar
MudSweatAndYears
Paddler
Posts: 27
Joined: May 24th, 2020, 6:31 am

Re: Compare rankings

Post by MudSweatAndYears » September 14th, 2020, 12:07 pm

sjors wrote:
September 10th, 2020, 1:08 pm

Some other interesting points you made, made me thinking: I.e. what about making age groups of 5 years. And just compare within that groups without compensating for age. Makes the whole thing very transparent and easy to code. 😀

Your last suggestion I don’t follow. Can you explain a bit more?
Using age groups of 5 years (of more) would work great provided you center these groups around the actual age of the rower. Taking you yourself (57 y, male, heavyweight) as an example: you should be competing against heavyweight males aged 55 - 59. Similarly, a 58 year old lightweight female should compete against 56 - 60 year old lightweight females.

To provide for competition across age, weight and gender categories one could define a ‘gold standard’ time for each category. Your ‘gold standard’ would be defined as the median of the fastest times for 55, 56, 57, 58 and 59 year old male heavyweights.

Perhaps a better alternative would be to compute the percentage of the rowers in your category who you beat. So in your case, if yourself included, 200 heavyweight males aged 55 - 59 compete, and you beat 180 of them (so 19 beat you), your score would be 180/200 = 0.90 or 90%. Note that this scoring algorithm suppresses anomalously high scores for groups with little or no competition: a 107 year old male lightweight who is the single competitor in the age bracket 105 - 109 and therefore competing against nobody, would end up with a score of 0%. Similarly, if there is 1 competitor (2 participants) in a category, the winner would score only 50%.

The score for each rower could be the gold standard time divided by the actual time achieved. (Slightly more refined would be to base the score on the Wattages: watts achieved divided by the watts for the gold standard.)

You will run into the issue of
I run in the mud, I sweat on the erg, and I happily battle the years...
M 59, 1.83m/6', 74kg/163lb. Aiming for sub-40 on 10k row and on 5k+5k row&run duathlon.

User avatar
MudSweatAndYears
Paddler
Posts: 27
Joined: May 24th, 2020, 6:31 am

Re: Compare rankings

Post by MudSweatAndYears » September 14th, 2020, 12:10 pm

MudSweatAndYears wrote:
September 14th, 2020, 12:07 pm
sjors wrote:
September 10th, 2020, 1:08 pm

Some other interesting points you made, made me thinking: I.e. what about making age groups of 5 years. And just compare within that groups without compensating for age. Makes the whole thing very transparent and easy to code. 😀
Using age groups of 5 years (of more) would work great provided you center these groups around the actual age of the rower. Taking you yourself (57 y, male, heavyweight) as an example: you should be competing against heavyweight males aged 55 - 59. Similarly, a 58 year old lightweight female should compete against 56 - 60 year old lightweight females.

To provide for competition across age, weight and gender categories one could define a ‘gold standard’ time for each category. Your ‘gold standard’ would be defined as the median of the fastest times for 55, 56, 57, 58 and 59 year old male heavyweights.

Perhaps a better alternative would be to compute the percentage of the rowers in your category who you beat. So in your case, if yourself included, 200 heavyweight males aged 55 - 59 compete, and you beat 180 of them (so 19 beat you), your score would be 180/200 = 0.90 or 90%. Note that this scoring algorithm suppresses anomalously high scores for groups with little or no competition: a 107 year old male lightweight who is the single competitor in the age bracket 105 - 109 and therefore competing against nobody, would end up with a score of 0%. Similarly, if there is 1 competitor (2 participants) in a category, the winner would score only 50%.
I run in the mud, I sweat on the erg, and I happily battle the years...
M 59, 1.83m/6', 74kg/163lb. Aiming for sub-40 on 10k row and on 5k+5k row&run duathlon.

Post Reply