Jump to content


Photo

History of Player Position Index (HOPPI)


  • Please log in to reply
35 replies to this topic

#21 The Prof

The Prof

    Colonel

  • NASF Committee
  • 1,515 posts
  • Coat of arms
  • Platinum Scout

Posted 30 June 2018 - 06:53 PM

When there is a tie, I think the scores should be the average of the scores for each of the tied positions.  For example, in a tie between 2nd and 3rd place in a 20-player tournament, each would receive (95+90)/2 = 92.5 points.  Or in a four-way tie for 4th place in a 50-person tournament, each would receive (94+92+90+88)/4 = 91 points.  This way, the same number of total points will be awarded for a tournament of a given number of participants regardless of how many ties there are. 


  • GaryLShelton and Fairway like this

#22 Wogomite

Wogomite

    Captain

  • NASF Committee
  • 857 posts
  • Coat of arms
  • Platinum Major

Posted 30 June 2018 - 08:18 PM

.

#23 Wogomite

Wogomite

    Captain

  • NASF Committee
  • 857 posts
  • Coat of arms
  • Platinum Major

Posted 30 June 2018 - 08:32 PM

.

#24 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 02 July 2018 - 11:06 AM

When there is a tie, I think the scores should be the average of the scores for each of the tied positions. For example, in a tie between 2nd and 3rd place in a 20-player tournament, each would receive (95+90)/2 = 92.5 points. Or in a four-way tie for 4th place in a 50-person tournament, each would receive (94+92+90+88)/4 = 91 points. This way, the same number of total points will be awarded for a tournament of a given number of participants regardless of how many ties there are.


It was only laziness that stopped me from doing this, but you're right - this method should be used. Will incorporate into the next update.

#25 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 04 July 2018 - 02:54 PM

Prof - a question, and another question.

Example:

Players finishing T9th in the Champions League (there are 8 of them). Currently they are assigned 9 as Finishing Value.

Under your proposal they would be assigned 12.5. Is this a fair reflection?

Example 2: In WCO many players finish on the same points, but different Buch.

Is it fair to give them a Finishing Value in order of Buch, or should Buch be discounted and they all finish Tied?

#26 The Prof

The Prof

    Colonel

  • NASF Committee
  • 1,515 posts
  • Coat of arms
  • Platinum Scout

Posted 10 July 2018 - 10:56 PM

Yes, I do feel 12.5 is a fair reflection for players tied for positions 9 through 16.  For a 32 person single elimination tournament things would work out like this: 1, 2, 3.5, 3.5, 6.5, 6.5, 6.5, 6.5, 12.5, …, 12.5, 24.5, …, 24.5.  As opposed to: 1, 2, 3, 3, 5, 5, 5, 5, 9,…, 9, 17,…, 17.  When this one is converted into a 1 to 100 scale, all those tied for 17th place get 50 points. (100 – 16*(100/32)), which seems too generous a result for players who don’t win a game.  In the first case, where they are considered tied for 24.5th place, they would get 26.6 points (100 – 23.5*(100/32))

 

In a league style, or Swiss system tournament, there are many fewer ties between places, so you don’t have this issue.  Here I think it’s fine to use Buch points to break ties.  More Buch points mean the player competed against stronger competition.  Had he had easier matchups he may have had more wins, so Buch points are a valid way to break ties.

 

However, I do see a problem with the Hoppi scoring system in general:  It makes the difference between any two consecutive places the same.  The difference between 1st and 2nd is the same as between 31st and 32nd; similarly the difference between 1st and 9th place, the same as the difference between 21st and 29th.  I think all would agree there should be a premium for top finishers.  The old system of TRP took this into account. 

 

In order to address this and yet keep the 1 to 100 system, I would suggest a modification:  Compute the scores for each position the current way and call this value x for each player.  Then compute y = x^2/100, or x squared divided by 100.  Then use each players y value as their Hoppi score.  In a tournament will 20 players, the x values are 100, 95, 90, 85, …, 5, but the y values for the various place values would be (1) 100, (2) 92 , (3) 81, (4) 72, (5) 64, (6) 56, (7) 49, (8) 42, (9) 36, (10) 30, (11) 25, (12) 20, (13) 16, (14) 12, (15) 9, (16) 6, (17) 4, (18) 2, (19) 1, (20) 0.  The model could be adjusted if you think it goes too far or not far enough toward compressing scores for the bottom half of the finishers.



#27 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 11 July 2018 - 08:32 AM

Thanks Prof.

Nice idea for the skewing of values towards the top. I suppose the difference between 31 and 32 shouldn't be vastly different between 14 and 15, but can see how the gap between places should grow at a quicker rate as you move into say the top 10 places.

One of the other areas that HOPPI doesn't consider is the opponent strength. It accounts for this indirectly - as you can assume that on average opponent strength increases as players progress through the tournament.

But I think it would be beneficial to get some measure of the playing field (in the form of + or - X) which you could apply to the Y values. How would you suggest doing this?

Suppose you took the average (mean - or possibly median) starting HOPPI score of every player who entered the tournament, and that used that to calculate a value between 0 and 5.

(Eg The average starting HOPPI scores have the below effect:

ASH = 65, Effect + 3
ASH = 60, Effect + 2
ASH = 55, Effect +1
ASH = 45, Effect -1
ASH = 40, Effect -2
ASH = 35, Effect -3 )

Do you think that would improve the system? This way eg Apex Division would carry more weight. (Although the I don't think coming 12th/12th in Apex should only earn a finishing value of 8, so perhaps there should be a minimum and maximum finishing value for each tournament depending on the field).

#28 The Prof

The Prof

    Colonel

  • NASF Committee
  • 1,515 posts
  • Coat of arms
  • Platinum Scout

Posted 11 July 2018 - 06:00 PM

It absolutely would be an improvement to take into account strength of opponents for one's tournament score.  But instead of trying to calibrate points to add or subtract, which is just guesswork, I'd recommend simply using average starting Hoppi of opponents to break any ties.  So now those 8 players that tie for 9th through 16th, can be assigned a specific position based on their opponents' starting Hoppi scores.  

 

Regarding the Pyramid League.  I agree the various divisions should not be considered in isolation, and that a minimum and maximum finishing value is a good idea.  I’d suggest that the range of scores given for Apex be from 100 down to the lowest starting Hoppi score of the 12 participants.  Core divisions could then have a range from the lowest starting Hoppi to the highest starting Hoppi, and Base a range from 1 to the highest starting Hoppi of the players in that division.



#29 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 11 July 2018 - 06:36 PM

In regards to your first point:

So effectively HOPPI starting score would operate similarly to Buch points. For players who didn't yet have a HOPPI score, I suppose the average starting score of all participants could be used.

I was considering the use of back-dating HOPPI to replace the score that an unrated player is awarded once they are ranked.

To explain more clearly, let's say that a player without a HOPPI has a default starting score of the average starting HOPPI. Let's assume it is 50. After 3 tournaments that player does well and now has a HOPPI score of 75. That score of 75 would be backdated to the previous 3 tournaments as if he had begun with 75.do you see any issues with this?

In regards to your second paragraph, I wonder if the minimum and maximum should extend beyond the upper and lower ranges of starting scores by a certain amount.

Eg Core division contains 12 players from 85 to 55. The player who wins would only get 85, but perhaps they should get a little more. Same applies for the player finishing last. (Eg Max value = highest HOPPI starting score plus 10%, min value = lowest HOPPI starting score less 10%)?

#30 The Prof

The Prof

    Colonel

  • NASF Committee
  • 1,515 posts
  • Coat of arms
  • Platinum Scout

Posted 12 July 2018 - 12:49 AM

If one of a player's opponents does not have a Hoppi score, then this opponent could be excluded in the calculation of the average.  This would only be a problem if all of a players opponents don't have a Hoppi score, which I don't think is a very likely scenario (Does the TC still run any single elimination tournaments?).  If you want to be sure to avoid this though then your idea of using the average starting score of all participants seems fine.

 

I also think extending the ranges for the various divisions by 10% to 20% is good.  It gives players a bit more to play for.  I don't think players are going to try to get in a lower division just to benefit their Hoppi score.

 

With regards to back-dating, this is problematic if you are using starting Hoppi to break ties.  After back-dating, some of a player's opponents may have had their starting Hoppi scores rise or fall, which could lead to a different result for the tournament than the original results.  This then affects the opponents strength numbers for the next tournament, and so on, and you can end up with many players having different current Hoppi scores.  Then do you use the new scores and back-date those?  What results is an iterative loop.  After many iterations the scores for each player likely converge to particular values, but this would be way too complicated a system.  



#31 Luckypapa

Luckypapa

    Lieutenant

  • Members
  • PipPipPipPipPipPipPip
  • 738 posts
  • Coat of arms
  • Gold Major

Posted 12 July 2018 - 07:14 PM

In 2013 I played in ‘The first online stratego tournament’ organized by Karaiskakis, starting october 1.
Why are the results of that great tournament - I beat Losermaker and Manning2Cruz before I lost in the semifinal to Nortrom - not in the table?

LuckyPapa

The secret of happiness is not in doing what you like, but in liking what you should do.


#32 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 12 July 2018 - 07:57 PM

In 2013 I played in ‘The first online stratego tournament’ organized by Karaiskakis, starting october 1.
Why are the results of that great tournament - I beat Losermaker and Manning2Cruz before I lost in the semifinal to Nortrom - not in the table?

LuckyPapa

 

http://forum.strateg...official online

 

Where are the results for this tournament?



#33 Major Nelson

Major Nelson

    Major

  • Moderators
  • 1,141 posts
  • Coat of arms
  • Platinum Marshal

Posted 12 July 2018 - 08:03 PM

When can we expect PRO Perfect to be added?

Winning isn't everything, but wanting to win is.


#34 Luckypapa

Luckypapa

    Lieutenant

  • Members
  • PipPipPipPipPipPipPip
  • 738 posts
  • Coat of arms
  • Gold Major

Posted 12 July 2018 - 08:18 PM

http://forum.strateg...official online

 

Where are the results for this tournament?

 

You can find it here:

http://forum.strateg...g-and-results/

 

 

My one moment of glory (or two if you want  :lol: ).

 

LuckyPapa 


Edited by Luckypapa, 12 July 2018 - 08:20 PM.

The secret of happiness is not in doing what you like, but in liking what you should do.


#35 TheOptician

TheOptician

    Marshal

  • Tournament Manager
  • 3,095 posts
  • Coat of arms
  • Platinum Captain

Posted 12 July 2018 - 09:49 PM

HOPPI Top 60 as of July 1st after PROPerfect

yn4k7im.jpg

HOPPI as of July 1st after PROPerfect

WG1Tiz9.jpg

Notes:

There are a number of changes to make that have not been incorporated in this update that hope to be featured in the next update and improve the system.

They include:


1. Breaking ties within a tournament by using average starting HOPPI score of each opponent - only where ties have not already been resolved by BUCH.

eg where four players tie for 5th currently '5' is used for each player to calculate Finishing Value. Instead, the players will be ordered 5 to 8 based on the strength of their opponents (calculated by their opponent's starting HOPPI score) that they faced within the tournament

2. Adjusting the range of Finishing Values available for each tournament based on the strength of the field (calculated by starting HOPPI scores)

ie Not all tournaments will have a maximum Finishing Value of 100 and a minimum Finishing Value of 0. Relatively weaker fields will carry a lower maximum, and relatively stronger fields will carry a higher minimum. The maximum Finishing Value of any tournament will be limited to 10% higher than the HOPPI starting score of the highest ranked player (provided this is not above 100). The minimum Finishing Value of any tournament will be limited to 10% lower than the HOPPI starting score of the lowest ranked player. 'Weaker' fields will therefore have a Finishing Value range of eg 30-90 as opposed to 0-100.

3. Divisions tournaments - which are currently treated as 'one' tournament, will instead be treated as individual tournaments. This will likely mean that players finishing top of eg Core will score better in that tournament than players coming eg bottom of Apex.

 

Special thanks to TheProf for his suggestions and advice in the past week. No doubt there will be more to come once the methodology of the above changes is presented :-)



#36 Wogomite

Wogomite

    Captain

  • NASF Committee
  • 857 posts
  • Coat of arms
  • Platinum Major

Posted 12 July 2018 - 11:33 PM

In 2013 I played in ‘The first online stratego tournament’ organized by Karaiskakis, starting october 1.
Why are the results of that great tournament - I beat Losermaker and Manning2Cruz before I lost in the semifinal to Nortrom - not in the table?

LuckyPapa

Lucky, very nice results. Well done. 

 

I see Nortrom won the first organized tournament at stratego.com, I am not surprised.


  • Luckypapa likes this




0 user(s) are reading this topic

0 members, 0 guests, 0 anonymous users