Swiss Tournament Edition IX.

[Sergio Martinez]

Conditions:

Swiss system tournament. The engines will be divided into groups of 80 engines + 20 best engines of the lower group . There will be five groups (A, B, C, D, E) and a last group (F), where new engines will be added. The sixteen best engines of group F will automatically promote to group E and will play a Swiss system against the 80 engines (12 rounds) that are in that group. Of those 100 engines, they will promote the first 20 to group D, and so on. All this in a single edition, so a new engine has the possibility to reach group A in that same edition. The 20 worst engines in each group will automatically demote.

Due to the number of new (very strong) participants I have decided that in this edition the number of engines that promote and demote will be increased to 20.

GPU: RTX 2080
GUI : Arena (3.51)
Time control: 40/4 (adapted to the CCRL benchmark).
Ponder: off
Threads : 4 New
Default hash=256Mb
Endgame tablebases: Syzygy - Nalimov - Scorpio Bitbases - Gaviota. Hash set to 64Mb.
Resign is set to -5 for 3 consecutive moves where able.
Updates will not be allowed during each tournament (each group).
Open to all Winboard and UCI engines (free, private and commercial).
The similarity tool will be used to determine whether or not an engine can participate (Engines allowed until 70% of moves match).
Results will be rated only for engines already in the CCRL lists.
All engines will use the same opening book limited to an 8 move depth.
Engine authors can either post their wish for their engine to participate or contact through email (sergiomg24@hotmail.com) or private message. I will choose engines by myself too.


Deadline to update engines: December 17
Start: December 18

[koedem]

Conditions:
[...] The sixteen best engines of group F will automatically promote to group E and will play a Swiss system against the 80 engines (12 rounds) that are in that group. [...]

I'm assuming that should be a twenty too?

If one wants to send a newer engine version can one request those games not being rated as to not skew the rating list? Or should one make an official release and then the games count towards the rating of that version? (but that wouldn't really work I guess, as one then would not have an initial rating; unless those tournament games are later added to the engine testing game pool)

[Sergio Martinez]

I'm assuming that should be a twenty too?

Yes, it was a writing error. Thank you

If one wants to send a newer engine version can one request those games not being rated as to not skew the rating list? Or should one make an official release and then the games count towards the rating of that version? (but that wouldn't really work I guess, as one then would not have an initial rating; unless those tournament games are later added to the engine testing game pool)

The rating of my tournaments has two origins:

-The first, as is obvious, the CCRL list. All engines that are included in my torunaments will take the rating of the version that is tested by CCRL. Otherwise, that is, that version of that engine has not been tested, either because it is a private version or because it has not been tested yet, I will use the rating of the version that has been tested.

-The second, my own estimates or using my lists. In these cases they are usually private engines. If it are new engines to me and I have not had time to test them, I make a rating estimate and assign it to them. If I could test them, I assign them the rating of my own lists.

When I start the tournament I use an engine version, if it's a private engine or a private version of that engine, it don't usually count in CCRL. If it is a freeware or commercial version, compute for CCRL, but always with the minimum of 300 games that we limit. If I use several versions during the tournament I mkeep the initial rating, in case of being commercial freeware versions and provided there is not a more recent version tested in CCRL.

Best regards and thanks for yout interest in my tournaments

[Sergio Martinez]

New entries for Edition IX (provisional):

Code: Select all

Allie
Black Diamond
Chess Fighter II
Fat Fritz
Fritz
Leelenstein
Stoofvlees

[Alexander Lim]

New entries for Edition IX (provisional):

Code: Select all

Allie
Black Diamond
Chess Fighter II
Fat Fritz
Fritz
Leelenstein
Stoofvlees

That looks tough. I wouldn't want to be in the bottom division with all those NN's!

[Alexander Lim]

Hi Sergio,

Time control: 40/4 (adapted to the CCRL benchmark).

Does this mean you will use a different time control on your computer? If so, what will it be?

Updates will not be allowed during each tournament (each group).

Does this mean updates are allowed if the engine promotes?

Best regards
Alexander

[xr_a_y]

Hi,

Minic 1.19 will be glad to participate !

Regards

[koedem]

When I start the tournament I use an engine version, if it's a private engine or a private version of that engine, it don't usually count in CCRL. If it is a freeware or commercial version, compute for CCRL, but always with the minimum of 300 games that we limit. If I use several versions during the tournament I keep the initial rating, in case of being commercial freeware versions and provided there is not a more recent version tested in CCRL.

Just to understand this correctly, if I submit a development version the results will not count to the existing versions rating?

[Sergio Martinez]

Hi,

Minic 1.19 will be glad to participate !

Regards

I count with Minic. It will play in Group B

[Sergio Martinez]

When I start the tournament I use an engine version, if it's a private engine or a private version of that engine, it don't usually count in CCRL. If it is a freeware or commercial version, compute for CCRL, but always with the minimum of 300 games that we limit. If I use several versions during the tournament I keep the initial rating, in case of being commercial freeware versions and provided there is not a more recent version tested in CCRL.

Just to understand this correctly, if I submit a development version the results will not count to the existing versions rating?

Exactly, not in CCRL list, but if I see a big improvement in results here, I will adjust the engine rating in my tournament.

The rating in this tournament is used to order the engines and thus make the pairings.

It also help to give us an idea of the strength of each engine (basically for new or private engines, the rest can be followed in some lists such as CCRL)

When the tournament finish, I upload the results to CCRL of the freeware or commercial engines. The games of the experimental versions, if available, also usually upload them, but it is normal that since they are not stable versions do not run more games (unless it is a long time before a new version and that version is finally stable).

Anyway, few games are played here, the tests for CCRL are more extensive and are the ones that really count for the list, even if these games are so few that there would be no difference in the CCRL rating although submit it.