Ordo 1.0.9 (new features for testers)

[Frank Quisinsky]

Code: Select all

   # PLAYER                           : RATING  PLAYED     (%)    D(%)  ERROR  OppAvg  OppErr    OppN  OppDiv
   1 Komodo 9.3 x64                   : 3179.1    2150   83.9%    28.1   15.3  2858.4     9.1      42    41.6
  76 Gaviota 1.0 AVX x64              : 2729.7    7150   42.3%    37.4    6.1  2799.7    11.1     120   113.2

Nice Miguel !!

Example:
Komodo have 100 games vs. Stockfish vs. all others the 50 games.
Gaviota have 100 or 150 games vs. a small number of opponents.

Good idea!

In my opinion:
The higher the number of games and the higher the number of opponents and the lesser the error of opponents the more exactly will be the own / final result in Elo.

That is the point!

I will view that in my next versions of my Rating list and will wrote a message for my news about it.

Best
Frank

Better is this one:

Code: Select all

   # PLAYER                           : RATING  PLAYED     (%)    D(%)  ERROR  OppAvg  OppDiv  OppErr
   1 Komodo 9.3 x64                   : 3179.1    2150   83.9%    28.1   15.3  2858.4    41.6     9.1

After draw quote the move average would be nice to see.

[Frank Quisinsky]

Move average:

Means:
For 1:0 for 0:1 and for 0.5:0.5 the move average.

Made sense for one eng-eng match or if People have a database of eng-eng games without resign mode. So we can see in additional nice information we can used for further statistics.

Best
Frank

[Ajedrecista]

Hello Miguel:

I added two of your suggestions
12: OppErr Average of the opponent errors
13: OppN Number of opponents

Which inspire me to add one that I think it is better than OppN

14: OppDiv

Diversity of opponents. If the number of games is equally distributed, it is the same as OppN, but if most of the games come from one opponent, it will get closer to 1. This is the "Effective number of opponents".

Code: Select all

   # PLAYER                           : RATING  PLAYED     (%)    D(%)  ERROR  OppAvg  OppErr    OppN  OppDiv 
   1 Komodo 9.3 x64                   : 3179.1    2150   83.9%    28.1   15.3  2858.4     9.1      42    41.6
  76 Gaviota 1.0 AVX x64              : 2729.7    7150   42.3%    37.4    6.1  2799.7    11.1     120   113.2

Example:
Komodo have 100 games vs. Stockfish vs. all others the 50 games.

What is the exact formula that you use for OppDiv? I find this option quite useful but I could not reproduce the value obtained by Frank. I tried the formulae from the following link:

Effective number of parties

Code: Select all

N_A = 1/SUM[(p_i)²]   (i = 1, ..., n)

N_B = SUM{(p_i)/[p_i + (p_1)² - (p_i)²]} = SUM(1/{1 + [(p_1)²/p_i] - p_i})   (i = 1, ..., n)

With p_1 >= p_2 >= ... >= p_n

In this example: n = 42; p_1 = 100/2150, p_2 = p_3 = ... = p_42 = 50/2150. I get N_A ~ 41.09 and N_B ~ 39.33 if I am not wrong. Thanks in advance.

Regards from Spain.

Ajedrecista.

[IWB]

12: OppErr Average of the opponent errors

What is OppErr good for and how do you define that?

8, 9 Errors of what?

Code: Select all

# PLAYER              : RATING  ERROR     (%)    D(%)  OppAvg   CFS(next)    POINTS       W       D       L  PLAYED  OppErr    OppN  OppDiv
   1 Stockfish 7         :   3239     10   80.1%    35.6    2978     100        2642.0    2054    1176      70    3300       8      15    15.0
   2 Komodo 9.2          :   3210     10   77.2%    33.5    2980     100        2548.5    1995    1107     198    3300       8      15    15.0
   3 Houdini 4           :   3124      9   67.5%    36.4    2985     100        2227.0    1626    1202     472    3300       9      15    15.0
   4 Gull 3              :   3071      9   60.8%    44.5    2989     100        2007.0    1272    1470     558    3300       9      15    15.0
   5 Ginkgo 1.5          :   3015      8   53.3%    49.6    2993      98        1758.0     939    1638     723    3300       9      15    15.0
   6 Equinox 3.30        :   3003      8   51.6%    49.3    2994      56        1702.5     889    1627     784    3300       9      15    15.0
   7 Fritz 15            :   3002      8   51.5%    48.9    2994      92        1698.5     891    1615     794    3300       9      15    15.0
   8 Critter 1.6a        :   2993      8   50.3%    49.6    2994     100        1659.0     841    1636     823    3300       9      15    15.0
   9 Protector 1.9.0     :   2944      8   43.6%    47.8    2997     100        1439.0     651    1576    1073    3300       9      15    15.0
  10 Nirvanachess 2.2    :   2926      9   41.1%    46.3    2999      90        1357.5     594    1527    1179    3300       9      15    15.0
  11 Andscacs 0.84       :   2918      8   40.0%    43.5    2999      93        1321.5     603    1437    1260    3300       9      15    15.0
  12 iCE 3.0             :   2909      9   38.8%    42.2    3000      88        1280.5     585    1391    1324    3300       9      15    15.0
  13 Jonny 7.01          :   2902      8   37.8%    40.8    3000      87        1249.0     575    1348    1377    3300       9      15    15.0
  14 Texel 1.05          :   2895      8   37.0%    41.3    3001      90        1220.0     539    1362    1399    3300       9      15    15.0
  15 Chiron 2            :   2887      8   36.0%    43.9    3001     100        1187.0     462    1450    1388    3300       9      15    15.0
  16 Naum 4.6            :   2867      9   33.4%    41.1    3003     ---        1103.0     425    1356    1519    3300       9      15    15.0

The other two new options are useless in my specific case

Thx for the update
Ingo

[Ozymandias]

12: OppErr Average of the opponent errors

What is OppErr good for and how do you define that?

8, 9 Errors of what?

That's the opponents' ELO error bar (weighted average?).

[IWB]

That's the opponents' ELO error bar (weighted average?).

Ahh, thx but no use for me then.

Anyhow, I like ordo and its possibilities

Ingo

[Frank Quisinsky]

Hi Miguel,

changed my conditions, worte News 265 and changed my FCP Rating List.

News 265:
http://www.amateurschach.de/main/_news.htm

FCP Rating List:
http://www.amateurschach.de/fcp-rating-list.txt
now with Ordo 1.0.9.5

I delete blank space and changed a bit to this one:

Code: Select all

  # Player                           : Elo     Games  Score% Draw%  Error  OppAvg OppDiv OppErr

Best
Frank

[michiguel]

12: OppErr Average of the opponent errors

What is OppErr good for and how do you define that?

8, 9 Errors of what?

That's the opponents' ELO error bar (weighted average?).

Correct.

Miguel

[michiguel]

Hello Miguel:

I added two of your suggestions
12: OppErr Average of the opponent errors
13: OppN Number of opponents

Which inspire me to add one that I think it is better than OppN

14: OppDiv

Diversity of opponents. If the number of games is equally distributed, it is the same as OppN, but if most of the games come from one opponent, it will get closer to 1. This is the "Effective number of opponents".

Code: Select all

   # PLAYER                           : RATING  PLAYED     (%)    D(%)  ERROR  OppAvg  OppErr    OppN  OppDiv 
   1 Komodo 9.3 x64                   : 3179.1    2150   83.9%    28.1   15.3  2858.4     9.1      42    41.6
  76 Gaviota 1.0 AVX x64              : 2729.7    7150   42.3%    37.4    6.1  2799.7    11.1     120   113.2

Example:
Komodo have 100 games vs. Stockfish vs. all others the 50 games.

What is the exact formula that you use for OppDiv? I find this option quite useful but I could not reproduce the value obtained by Frank. I tried the formulae from the following link:

Effective number of parties

Code: Select all

N_A = 1/SUM[(p_i)²]   (i = 1, ..., n)

N_B = SUM{(p_i)/[p_i + (p_1)² - (p_i)²]} = SUM(1/{1 + [(p_1)²/p_i] - p_i})   (i = 1, ..., n)

With p_1 >= p_2 >= ... >= p_n

In this example: n = 42; p_1 = 100/2150, p_2 = p_3 = ... = p_42 = 50/2150. I get N_A ~ 41.09 and N_B ~ 39.33 if I am not wrong. Thanks in advance.

Regards from Spain.

Ajedrecista.

Suppose a player played against 3 opponents (1,2,3) the following amount of games: n1, n2, and n3.

(total games) N = n1 + n2 + n3

fraction of games played
f1 = n1/N
f2 = n2/N
f3 = n3/N

Then the following summation:

I = - (f1 * log(f1) + f2 * log(f2) + f3 * log(f3))

I is the Shannon information content and log is natural logarithm

OppDiv = exp (I)

Miguel

[michiguel]

13: OppN Number of opponents

Which inspire me to add one that I think it is better than OppN

14: OppDiv

Diversity of opponents. If the number of games is equally distributed, it is the same as OppN, but if most of the games come from one opponent, it will get closer to 1. This is the "Effective number of opponents".

That seems more in the spirit of the request.

Now for mine, regarding players with few games. It's a bit inconvenient, that you have to use the -g switch when some players are under-represented. It should be possible to specify the minimum number of games for a player to be rated, and after the program runs, Ordo could list those that were left out and how many games they had. In a related request, I would also include an option for the user to specify the percentage of games a cluster needs, to be rated. Below that mark, I wouldn't rate games from marginal clusters.

ordoprep can clean up a file removing players setting some limits.
https://github.com/michiguel/Ordoprep

I may include those options directly into ordo at one point.
Miguel