CEGT League System + World Trophy (deadline postponed)

[Heinz van Kempen]

Hi all ,

needing more suspense I am proposing to the CEGT testers to start CEGT leagues now with the additional hardware.

My suggestions will be:

-three classes with 20 engines each and CEGT time control (40/40 2Ghz), four demoting and promoting. Rank 15 and 16 will get playoff matches against position 5 and 6 from the lower class for being promoted/demoted. Alternatively these engines play a mini tournament for promotion/demotion. Classes will be build according to the CEGT rating list (standing Saturday night). One version for each engine (an exception might be Fruit and Toga)

-in the highest class we will include the best Chessmaster setting recommended by the experts (Graham and Ray Banks, Jason Kent etc.) and either Fruit or Toga or both (depending on votes of testers)

-updates are possible at every time within the tournaments when the engine version replaced already has more than 500 games in CEGT rating list

- any GUI can be used when possible (depending on participants)

-I will calculate how many tournaments will be run. Each cycle (including all three classes and a promotion tournament for the third class) should last three or four months. All classes are run simultaneously.

Engine authors who want to replace their engine (when it already has more than 500 games in the list) please write to:

hvankempen@web.de

(betas can be included if the majority of testers agrees)


Additional testers as always are welcome to help. It will surely be a great event with statistical significance.

Best Regards
Heinz

[Graham Banks]

Hi Heinz,

Ray and I would recommend that CM10th Cell (Dom Leste's setting) be used this time round.
Cell has the best performance to date in Ray's testing (consisting of each setting playing 280 games against other top programs).
Cell is also near the top of the CM10th Showdown.
Not sure if Jason concurs, but I notice that Cell has performed well in Jason's testing against Shredder 9 also.
I conclude therefore that at this point in time, it is the most consistently well performed CM10th setting.

Regards, Graham.

[Heinz van Kempen]

Hi Graham ,

thanks for your proposal. Cell will be taken if there will be no other votes.

From opinions I received from testers so far we will include Fruit and Toga.

Best Regards
Heinz

[Heinz van Kempen]

Hi all ,

based on the current CEGT rating list the classes and testers would be the following (versions may change until deadline Saturday evening 9 p.m. and also qualify can change when the rating list differs at this time).

The tournaments should run three months and give several hundred games for each engine:

Master Class (classes will be probably renamend):

Michael
Rainer
Charles 1

Code: Select all


  1 Shredder 9                     : 2750   11  11  2990    68.4 %   2616   30.8 %
  2 Toga II 1.0                    : 2722   22  22   632    60.2 %   2650   36.2 %
  3 Fruit 2.1                      : 2712   12  12  2258    61.2 %   2632   34.8 %
  4 Fritz 8 Bilbao                 : 2707   13  13  1946    60.3 %   2635   30.3 %
  5 Junior 9                       : 2690   12  12  2129    59.4 %   2624   31.8 %
  6 Hiarcs 9                       : 2667   11  11  2382    55.3 %   2630   35.7 %
  7 Gandalf 6.0                    : 2658   11  11  2462    54.9 %   2624   34.3 %
  8 Ktulu 7.0a                 : 2658   13  13  1881    57.0 %   2609   33.1 %
  9 Chess Tiger 15.0               : 2652   11  11  2223    53.7 %   2626   38.1 %
 10 Spike 1.0a Mainz               : 2651   26  26   479    58.4 %   2593   31.5 %
 11 Slow Chess Blitz WV2           : 2649   39  39   216    54.9 %   2615   30.1 %
 12 CM 10000 (Cell)             : 2638   18  18   915    50.3 %   2636   37.5 %
 13 Ruffian 2.1.0                  : 2638   13  13  1729    52.3 %   2622   34.3 %
14 List 512                       : 2630   14  14  1519    49.3 %   2635   33.6 %
15 Pro Deo 1.1                    : 2629   15  15  1424    49.5 %   2633   30.6 %
16 SOS 5 for Arena                : 2601   15  15  1398    45.6 %   2631   35.3 %
17 Pseudo 0.7c                    : 2598   16  16  1113    48.3 %   2609   36.8 %
18 Aristarch 4.50                 : 2598   14  14  1494    46.1 %   2625   32.8 %
19 Zappa 1.1                      : 2597   50  50   128    53.9 %   2570   32.8 %
20 Naum 1.8                       : 2595   24  24   515    53.9 %   2568   38.6 %



Higher Class

Uschi + Chris (combining)
Charles (second computer)
Heinz 1

Code: Select all


1 Pharaon 3.3                    : 2589   18  18  1010    54.1 %   2561   33.5 %
2 Glaurung Mainz                 : 2589   25  25   516    52.4 %   2572   28.9 %
3 Jonny 2.82                     : 2580   28  28   430    54.3 %   2550   28.6 %
4 Anaconda 2.0.1                 : 2574   15  15  1335    44.9 %   2610   34.2 %
5 Scorpio 1.3                    : 2573   20  20   747    52.3 %   2557   33.5 %
6 Little Goliath Evolution       : 2566   24  24   532    51.1 %   2558   36.1 %
7 WildCat 5                      : 2562   25  25   494    50.7 %   2557   31.8 %
8 Thinker 4.7a                   : 2558   15  15  1321    45.6 %   2588   38.4 %
9 Delfi 4.5                      : 2557   16  16  1320    42.6 %   2609   32.3 %
10 The Baron 1.7.0                : 2557   51  51   127    45.3 %   2590   29.1 %
11 Green Light Chess 3.01.2.2     : 2543   18  18   908    47.4 %   2561   35.1 %
12 AnMon5.53                      : 2535   21  21   777    49.0 %   2542   29.6 %
13 Zarkov 4.75                    : 2525   25  25   486    46.0 %   2553   35.6 %
14 Crafty Cito 1.2                : 2524   24  24   553    47.9 %   2538   34.4 %
15 Ufim 7.01                      : 2511   34  34   273    42.9 %   2561   31.5 %
16 Amyan 1.595                    : 2506   17  17  1142    36.4 %   2602   30.6 %
17 Trace 1.34                     : 2505   40  40   215    41.6 %   2564   28.4 %
18 Movei 00.8.310                 : 2501   18  18   950    41.3 %   2562   33.6 %
19 Yace 0.99.87                   : 2500   21  21   740    36.8 %   2594   31.6 %
20 Petir 2.75                     : 2496   24  24   503    42.5 %   2548   37.4 %



Challenger Class

Ralf
Heinz (second computer)
Heinz (third computer)

Code: Select all


1 King of Kings 2.56             : 2488   61  61    89    39.3 %   2564   31.5 %
2 ETChess 30.08.05               : 2487   28  28   414    40.1 %   2557   31.9 %
3 Pepito v1.59                   : 2485   36  36   232    40.1 %   2555   37.1 %
4 Bruja 1.9                      : 2434  124 130    28    33.9 %   2549   17.9 %



Here engines will be added when authors show interest to participate.

Email to hvankempen@web.de

If not the third class will be dropped and converted to a smaller promotion tournament adding some promising engines. Then there would be more tournaments in Master and Higher Class.

Best Regards
Heinz

[Heinz van Kempen]

Hi all ,

an interesting proposal came from Rainer. He likes to have three systems and three tournaments with 48 engines and then a combined point system to decide the best. We had something similar in the City Grand Prix.

1. 4 preliminary groups with 12 participants, 11 rounds
first three from each group play in the final, 11 rounds

2. 11 round Swiss

3. Knockout system, 4 round each

Would this be more interesting than a normal league system?

I gave this for voting to the CEGT testers. Seems that we feel a bit bored currently . The other question is if to continue with CEGT, as there is already the same problem we had with AEGT (lack of support).

Best Regards
Heinz

[Graham Banks]

Personally I prefer the league system. Knockout matches mean little unless there's a minimum of 20 to 30 games per match.

[Heinz van Kempen]

Hi Graham ,

thanks, the numbers here were just an example taken from Rainer?s mail. Of course with so many machines we could also run 40 games knockout matches each.

The problem of testing in a group is that you have a lot of different proposals and opinions and have to find a compromise so that everyone is happy and motivated. Not easy, believe me .

But I would be interested to hear also from others what system would best fit to CEGT.

Best Regards
Heinz

[Heinz van Kempen]

Hi all ,

as expected votings were neatly split and Rainer proposed the all in one, all inclusive solution. So at least eight computers will be used for the following:

- 2 x League System A and B

with less games then and 20 already mentioned engines for each group. I think we will then replace one "older" engine with E.T.Chess, to honour a newly released fine engine. There will be no group C for lack of interest from the authors, but maybe a promotion tournament later

- 1 x World Trophy (maybe it will be renamed) with 6 tournaments each as follows:

to give Uschi some work with background pictures we will run three different tournaments (knockout, round robin, Swiss) in each country for two groups of 16 engines, one class for the top engines including Fruit, Toga, CM10th Cell, Shredder, Spike, etc. and another for those where we still have few games in the rating list (proposals for the groups here are below).

The best in the three kinds of tournaments will get points, those will be added to an overall score and then the winner for each country is decided. Maybe we will give gold, silver and bronze medals and the engine winning the most gold medals will be the World Trophy winner.

Here is a proposal for the classes for the World Trophy:

Class A

Code: Select all


  1 Shredder 9                     : 2750   11  11  2990    68.4 %   2616   30.8 %
  2 Toga II 1.0                    : 2722   22  22   632    60.2 %   2650   36.2 %
  3 Fruit 2.1                      : 2712   12  12  2258    61.2 %   2632   34.8 %
  4 Fritz 8 Bilbao                 : 2707   13  13  1946    60.3 %   2635   30.3 %
  5 Junior 9                       : 2690   12  12  2129    59.4 %   2624   31.8 %
  6 Hiarcs 9                       : 2667   11  11  2382    55.3 %   2630   35.7 %
  7 Gandalf 6.0                    : 2658   11  11  2462    54.9 %   2624   34.3 %
  8 Ktulu 7.0a                 : 2658   13  13  1881    57.0 %   2609   33.1 %
  9 Chess Tiger 15.0               : 2652   11  11  2223    53.7 %   2626   38.1 %
 10 Spike 1.0a Mainz               : 2651   26  26   479    58.4 %   2593   31.5 %
 11 Slow Chess Blitz WV2           : 2649   39  39   216    54.9 %   2615   30.1 %
 12 CM 10000 (Cell)             : 2638   18  18   915    50.3 %   2636   37.5 %
 13 Ruffian 2.1.0                  : 2638   13  13  1729    52.3 %   2622   34.3 %
14 List 512                       : 2630   14  14  1519    49.3 %   2635   33.6 %
15 Pro Deo 1.1                    : 2629   15  15  1424    49.5 %   2633   30.6 %
16 SOS 5 for Arena                : 2601   15  15  1398    45.6 %   2631   35.3 %



Class B

Code: Select all


1 Zappa 1.1                      : 2597   50  50   128    53.9 %   2570   32.8 %
2 Naum 1.8                       : 2595   24  24   515    53.9 %   2568   38.6 %
3 Glaurung Mainz                 : 2589   25  25   516    52.4 %   2572   28.9 %
4 Jonny 2.82                     : 2580   28  28   430    54.3 %   2550   28.6 %
5 Scorpio 1.5y                    : 2573   20  20   747    52.3 %   2557   33.5 %
6 Little Goliath Evolution       : 2566   24  24   532    51.1 %   2558   36.1 %
7 WildCat 5                      : 2562   25  25   494    50.7 %   2557   31.8 %
8 Ufim 7.01                      : 2511   34  34   273    42.9 %   2561   31.5 %
9 Trace 1.34                     : 2505   40  40   215    41.6 %   2564   28.4 %
10 Petir 2.75                     : 2496   24  24   503    42.5 %   2548   37.4 %
11 ETChess 30.08.05               : 2487   28  28   414    40.1 %   2557   31.9 %

and five free spots still do decide



Hope that a few authors will show their support for CEGT and increase the motivation of the testers by giving us a fine beta or by releasing a new engine before Saturday night.

Best Regards
Heinz

[Heinz van Kempen]

Hi all ,

deadline for the League System and the World Trophy is postponed to Thursday evening.

CEGT testers are currently feverishly testing only one engine and so we have the opportunity to run several hundred games with CEGT time control before the tournaments keep us too busy.

Best Regards
Heinz