News:

Herr Otto Partz says you're all nothing but pipsqueaks!

Main Menu

SWR changes!

Started by Mark L. Rivers, December 18, 2006, 09:43:56 PM

Previous topic - Next topic

Mark L. Rivers

SWR changes![/b]

Beginning from 1st January 2007 actual rank will be re-calculated after the introduction of 3 important changes:

1) New calculation system will be introduct to avoid that low or high rates of participation affect abnormally the rank.

2) SWR will be calculated considering only final results of races (listed in SWR page) finished in last 8 months. Update becomes monthly.

3) For every driver, score got will be integrated with additional points comparing 1) his result with results of other drivers which have participated with him at the same race, and 2) respective positions in current SWR.


A full description of new SWR criteria will be published next week in this topic and contextually in SWR News page.


zaqrack

nice! sounds like a more balanced system to me!

Krys TOFF

Quote from: zaqrack
nice! sounds like a more balanced system to me!
True, but that means I'll be out of top 10... :-\

BonzaiJoe

They sound like good changes. Let's see how it works...
But we can't be quite sure.


Chulk

Quote from: Krys TOFF on December 19, 2006, 09:50:31 AM
Quote from: zaqrack
nice! sounds like a more balanced system to me!
True, but that means I'll be out of top 10... :-\
So will I I'm afraid
Yes, it is me. No, I'm not back at racing (for now...)

Mark L. Rivers

#5
Well, it's time to explain more detailedly how new SWR criteria will work beginning from 1st January 2007.

Recapitulating, there will be three important changes:

1) New calculation system
2) New Races Database
3) Results comparation with current ranking





1) New system of calculation

With actual criteria, Total Points are calculated summing Average Race Points and a Coefficient, that is Participation?s Rate to official races multiplied 10.
New criteria will be calculated multiplying Average Race Points with a new kind of coefficient. This is the formule of new coefficient: (Log12(Participation?s Rate) + 1,5)

How does new Coefficient work? Keeping on mind that Participation?s Rate is included between 0 and 1, see the graphic below:





As graphic shows, a Participation?s Rate lower than 1/3 (0,33) is very penalizing for a pipsqueak. So even if he reaches great results in 4 of 5 races in last period, he correctly wouldn?t reach an high position on SWR. But going on, the Coefficient?s curve decreases logarithmically his gradient, and this means that a very high Participation?s Rate doesn?t affect on Total Points too much respect a medium Participation?s Rate, as it happened with previous criteria.

Why Log12? And why add it 1,5? There?s no a complex and specific mathematic calculation behind new coefficient. Factors which have contribute to calculate it, have been choosen looking for a curve that could obtains the desired target, that is, as aforesaid, to avoid that a low or an high rate of participation affect abnormally the rank. After many trials I think to have individuate a curve which reaches this target with a right balance.

To explicate this concept, see the following examples:

Races in last 8 months = 30

Driver ?A? has raced 20 races with Avg Race Pts pair to 13,3 points.
His Participation? Rate is 0,667 and so his coefficient is 1,337
Multiplying his Avg Race Points with his coefficient he reaches 17,780 Total Points



Driver ?B? has race 8 races with Avg Race Points pair to 17,5 points
His Participation?s Rate is 0,267 and so his coefficient is 0,968
Multiplying his Avg Race Points with his coefficient he reaches 16,942 Total Points

Comments: Very high Avg Race Points doesn?t permit to stay ahead Driver ?A? because a low Participation?s Rate, which determines that Total Points are a bit lower than Avg Race Points.


Driver ?C? has raced 3 races with Avg Race Pts pair to 20 points (three victories).
His Participation? Rate is 0,100 and so his coefficient is 0,573
Multiplying his Avg Race Points with his coefficient he reaches 11,467 Total Points

Comments: A very, very low Participation?s Rate penalizes correctly results of Drivers ?C?. His Total Points are consistently lower than Avg Race Points.


Driver ?D? has race 30 races with Avg Race Points pair to 11 points
His Participation?s Rate is 1,000 and so his coefficient is 1,500
Multiplying his Avg Race Points with his coefficient he reaches 16,500 Total Points

Comments: Highest Participation?s Rate (all races competed) is not enough to stay ahead Driver ?A? (only 2/3 of races competed) even if respective Avg Race Points are quite near.




2) New Races Database

Races Database to calculate SWR will be reducted and it will include only final results of official races finished in last 8 months. SWR will be normally update the 5th day of every month. For every update, a new range of dates will be calculated to individuate which races constitute the official Races Database. Start date of every range will be the 5th day of the 8th previous month. For example, when SWR will be update the 5th March 2007, Races Database will include only official races finished after 5th July 2006. The list of official races included will be published in SWR page with all SWR re-calculated score.




3) Results comparation with current rank

Re-calculated exponential score used until is integrated with additional points comparing:

a) result of every single driver with results of other drivers which have participated with him at the same race, and 

b) respective positions in current SWR.

One point will be assigned (+1 point) to a driver for every driver which will result behind him in race classification and ahead him in SWR.
One point will be leaved (-1 point) to a driver for every driver which will result ahead him in race classification and behind him in SWR.

An example to see how this concept works:

SWR                       FINAL RESULTS OF A NEW RACE

Driver                   Driver       Score      Additional Points                                             TOT
01) A                      01) D        20        +3 (ahead C-A-B)         0                                 23
02) B                      02) C        15        +2 (ahead A-B)           -1 (behind D)                  16
03) C                      03) A        12          0                            -2 (behind D-C)               10
04) D                      04) B          8          0                            -2 (behind D-C)                6
05) E                      05) L          8        +5 (ahead G-H-F-I-E)    0                                 13
06) F                      06) G          8        +2 (ahead F-E)           -1 (behind L)                    9
07) G                      07) H          4        +2 (ahead F-E)           -1 (behind L)                    5
08) H                      08) F          4        +1 (ahead E)              -3 (behind L-G-H)              2
09) I                       09) I          4        +1 (ahead E)               -1 (behind L)                    4
10) L                       10) E         4          0                             -5 (behind L-G-H-F-I)       -1




Well, that's all.  ;)
Current rank will be soon re-calculated with new criteria.
I hope new SWR could be really more balanced and a pleasing component of Stunts racing world!  :)




Krys TOFF

#6
Quote from: Mark L. Riverssee the graphic below:
Mmmm, I don't see any graphic below. ??? :-\

Quote from: Mark L. Riversthe Coefficient?s curve decreases logarithmically his gradient, and this means that a very high Participation?s Rate doesn?t affect on Total Points too much respect a medium Participation?s Rate, as it happened with previous criteria.

Why Log12? And why add it 1,5? There?s no a complex and specific mathematic calculation behind new coefficient. Factors which have contribute to calculate it, have been choosen looking for a curve that could obtains the desired target, that is, as aforesaid, to avoid that a low or an high rate of participation affect abnormally the rank. After many trials I think to have individuate a curve which reaches this target with a right balance.
OK. But, as an example, could you tell us what would have been the ISM ranking of 2006 with this new criteria compared to the one used in 2006 ? It would be the best way to let everybody understand, isn't it ? ;D

Quote from: Mark L. RiversCurrent rank will be soon re-calculated with new criteria.
OK, just what I asked before then. You read my mind, didn't you ? ;D

zaqrack

nice job!
probably the most complicated Stunts ranking ever, but it sounds good!

Krys TOFF

#8
Quote from: Mark L. RiversAn example to see how this concept works:

SWR                       FINAL RESULTS OF A NEW RACE

Driver                   Driver       Score      Additional Points                                             TOT
01) A                      01) D        20        +3 (ahead C-A-B)         0                                 23
02) B                      02) C        15        +2 (ahead A-B)           -1 (behind D)                  16
03) C                      03) A        12          0                            -2 (behind D-C)               10
04) D                      04) B          8          0                            -2 (behind D-C)                6
05) E                      05) L          8        +5 (ahead G-H-F-I-E)    0                                 13
06) F                      06) G          8        +2 (ahead F-E)           -1 (behind L)                    9
07) G                      07) H          4        +2 (ahead F-E)           -1 (behind L)                    5
08) H                      08) F          4        +1 (ahead E)              -3 (behind L-G-H)              2
09) I                       09) I          4        +1 (ahead E)               -1 (behind L)                    4
10) L                       10) E         4          0                             -5 (behind L-G-H-F-I)       -1


One potential risk appears in my mind : some pipsqueaks would prefer not to send a replay for one race they know they would not reach a good result just to avoid getting negative points. As long as a pipsqueak race 2/3 of the tracks in a year is enough, so it's a risk.
Like it happened a few years ago when ZakStunts scoreboard was based on average ranking and not sum of points if I remember it well.

Quote from: zaqrack
nice job!
probably the most complicated Stunts ranking ever, but it sounds good!
I reminds me tennis points : if you won a tournament the year before, you can not earn more points if you win it again the year after, but you can loose some if you do not win it again.
Beeing 1st in this ranking will not be hard (well, for me it will, but I talk about top drivers), the main problem will be to remain 1st at this ISM ranking. One bad result and you'll loose numerous points.

Mark L. Rivers

#9
SWR has been re-calculated with new criteria!  :)

A little changes has been apported furtherly as new SWR criteria.
Additional points (renamed Extra Points) won't be summed at standard scoring (as aforesaid declared) but they will be summed at Average race Points. To balance new system of calculation, Extra Points won't have a unitary value (1, 2, 3 etc. or -1, -2, -3 etc.) but they will have a decimale value (0.1, 0.2, 0.3 etc. or -0.1, -0.2, -0.3 etc.).

Follow me while I try to explain better the thing.

Before this further modify the calculation was as the following example:

Average Race Points = 14,5
Standard Score = 8 points
Stand Points   = 3 points
Race Points    = 8 + 3 = 11 points  (in this way Extra Points would have affected Average Race Points)
New Avg Race Points = 14,1

Coefficient = 0,75
Total Points = New Avg Race Points multiplying Coefficient



After new modify, the claculation is as the following example:

Average Race Points = 14,5
Standard Score = 8 points
Race Points    = 8 points
New Avg Race Points = 13,9

Stand Points   = 0,3 points

Base Points = New Avg Race Points + Stand Points = 14,2
Coefficient = 0,75
Total Points = Base Points multiplying Coefficient


Final result doesn't change very much than before this further modify, but in this way statistics about Avg Race Points remain "clean" and all calculations will result more "trasparent" seeing SWR scoreboard.  ;)


Anyway you can see immediately new SWR!: digilander.libero.it\stunts.SDR\ISM\SWR.htm

Krys TOFF

With new system, I drop from 5th to 8th place. Good, I feared to be out of top ten.

Mark L. Rivers

Quote from: Krys TOFF on December 30, 2006, 09:30:59 PM
With new system, I drop from 5th to 8th place. Good, I feared to be out of top ten.

If I won't improve a lot to stay ahead some top driver at least in some race, I fear to drop very low with new Stand Points....   ::)

CTG

Too much suffer for a single list. I keep my own WRL settings.

Mark L. Rivers


Beginning from today SWR site changes address. New address is the following:

http://digilander.libero.it/stunts.SDR/SWR/SWR.htm

All links from ISM site have been modified yet.

Mark L. Rivers

#14
Two consistent modifies has been applied at SWR to make it more balanced!

1) New exponential score (Race Points)
2) Quality Race Index


Por la traduccion en espa?ol --> http://digilander.libero.it/stunts.SDR/SWR/SWR_criteria.htm

------------------------

1) NEW EXPONENTIAL SCORE (Race Points)

Last months races has showed a more homogeneous number of participants at the official races. The new exponential score attributed to every single race of every contest will permit:

A) a more suitable and balanced score
B) a specific number of points for the first 15 drivers of every race (differently from actual score where some drivers, for example 4th, 5th and 6th, get the same number of points)

So, Race Points of last 8 months (also Stand Points attributed yet) have been recalculated in this way:

60 points at 1st
50 points at 2nd
40 points at 3rd
32 points at 4th
26 points at 5th
22 points at 6th
18 points at 7th
15 points at 8th
12 points at 9th
10 points at 10th
  9 points at 11th
  8 points at 12th
  7 points at 13th
  6 points at 14th
  5 points at 15th
  2 points from 16th to 20th
  1 point from 21st to the last

A single Stand Point changes from actual 0,1 point to 0,3 point.


In SWR NEWS page (http://digilander.libero.it/stunts.SDR/SWR/SWR_news.htm) you can see a table reporting the percentage of Race Points gained by drivers placed from 2nd to the last comparing with Race Points gained by the winner. The comparison table includes new SWR score, previous SWR score and the score of active contests.

---------------------------

2) QUALITY RACE INDEX

Beginning from this update a new factor has been introduced.
As I said when SWR was created I don?t think correct to attribute different scores to the active contests.
As I said, I think that what really makes a contest more difficult than other one, it's not the number of participants in absolute (in some races, the number of participants is enlarged by some occasional driver which would occupy the same positions in every contest), and it's not the quality of the tracks (I really think that the level is high everywhere). This is the reason to lead me to say that a race is equal to other one, independently in which contest it's competed.
Anyway, it?s true that, attributing a unique score to every race, in some races it?s easier to get a high score than others because the quality of the participation is not very high. For example, I think that my victory at HUTSPOT (and my 60 points) has not the same value of my victory at ZALAEGERSZEG. 
So I thought to introduce in the criteria of calculation a new factor, keeping in count the quality of the race, that is the rank of the drivers that take part at the race.
Let me explain how new factor is calculated.

Every SWR update determines a new rank. To the first 25 drivers is attributed a specific position value in this way:

1st = 10
2nd = 9
3rd = 8
4th = 7
5th?7th = 6
8th?10th = 5
11th?15th =4
16th?20th =2
21st?25th =1

For every race it?s calculated an index of the quality of the race, that is:

Sum of the position values of each participant - 30


For example, drivers in ZALAEGERSZEG (finished the 14th mar 07) were:

Mark L. Rivers (4th in last SWR update, 5th Mar 2007, position value 7)
Dottore (5th in last SWR update - position value 6)
CTG (2nd in last SWR update - position value 9)
Werda (15th in last SWR update - position value 4)
CTG (11th in last SWR update - position value 4)
Krys TOFF (8h in last SWR update - position value 5)
Kub (24th in last SWR update - position value 1)
CTG (21st in last SWR update - position value 1)
SuperBrian (14th in last SWR update - position value 4)
Diesel Joe (22th in last SWR update - position value 1)
AbuRaf70 (13th in last SWR update - position value 4)
Youri (43th in last SWR update - no value)
Sergio Baro (28th in last SWR update - no value)

So the quality race index is:  46 - 30 = 16

Quality Race Points are attributed to the drivers in this way:

1st Mark L. Rivers ? Quality Race Points = 16/10/1 = 1,6
2nd Dottore              QRP = 16/10/2 = 0,8
3rd CTG                   QRP = 16/10/3 = 0,5
4th Werda                QRP = 16/10/4 = 0,4
5th CTG               QRP = 16/10/5 = 0,5
6th Krys TOFF         QRP = 16/10/6 = 0,4
7th Kub                    QRP = 16/10/7 = 0,3
8th CTG             QRP = 16/10/8 = 0,3
9th SuperBrian         QRP = 16/10/9 = 0,2
10th Diesel Joe        QRP = 16/10/10 = 0,2
11th AbuRaf70         QRP = 16/10/11 = 0,2
12th Youri                QRP = 16/10/12 = 0,2
13th Sergio Baro      QRP = 16/10/13 = 0,1

--------------------

Other example for HUTSPOT (finished the 28th Mar 07):

Mark L. Rivers (4th in last SWR update, 5th Mar 2007, value 7)
Krys TOFF (8h in last SWR update - value 5)
AbuRaf70 (13th in last SWR update - value 4)
Paleke (20th in last SWR update - value 2)
Hek (16th in last SWR update - value 2)

So the quality race index is:  20 - 30 = -10

Quality Race Points is attributed to the drivers in this way:

1st Mark L. Rivers ? Quality Race Points = -10/10/1 = -1,0
2nd Krys TOFF       QRP = -10/10/2 = -0,5
3rd AbuRaf70         QRP = -10/10/3 = -0,3
4th Youri                 QRP = -10/10/4 = -0,3
5th Sergio Baro      QRP = -10/10/5 = -0,2


As you can see, Quality Race Points can be positive or negative.

How QRP works in SWR criteria?

SWR criteria was calculated until last update in this way:

Base Points = Average Race Points + Stand Points
Coefficient = Log12(Participation Rate) ? 1,5
Total Points = Base Points x Coefficient

Beginning from this update, Base Points is calculated summing Quality Race Points too, so:

Base Points = Average Race Points + Stand Points + Quality Race Points

Calculation of Coefficient and Total Points remains the same.

So:

Quality Race Points doesn?t affect on Score gained (it remains the same for every contest), but
Quality Race Points affects sensibly Base Points decreasing the value to multiply the Coefficient


------------------------------------

Definitely, the criteria to determine how much a victory have to be rewarded is not

1) Number of participants (QRP is affected very marginally from some occasional participants) or
2) Prestige and historicity of a contest (even if there?s sure a particular pleasant to win a race in one of them),

but is the Quality Race Index, an index which can be high in a whatever race of a whatever contest.

---------------------

To make more transparent the change, this month in SWR list page you can see, beside the previous Rank (reference Rank for Stand Points and Quality Race Points), the new Rank without new changes and the new rank with new Exponential Score and new factor of calculation: Quality Race Points.

http://digilander.libero.it/stunts.SDR/SWR/SWR.htm