Important News: I decided to stop doing my SPCC-Ratinglist, because the balanced openings, used for this ratinglist, are not working anymore in these days of superstrong engines... My UHO-Ratinglist will take over here: I will add 5 more engines to this huge RoundRobin-tournament, so a new UHO-Top15 Ratinglist will come in 1 or 2 weeks from now. Because the measurement of Stockfish-Dev progress must be continued, the reference-Elo of this UHO-Top15 Ratinglist will no longer be 0 Elo for Stockfish 16, but 3821 Elo for Stockfish 16 (Elo of SF 16 in my SPCC-Ratinglist).
The EAS-Ratinglist will be calculated out of the UHO-Top15 Ratinglist games in the future.
If you want to download the games of the SPCC-Ratinglist, do it right now, because in 6-10 days, the SPCC-Ratinglist will be removed from my website.
SPCC: Classical Ratinglist will be stopped
Moderator: Ras
-
pohl4711
- Posts: 2918
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
-
Jouni
- Posts: 3840
- Joined: Wed Mar 08, 2006 8:15 pm
- Full name: Jouni Uski
. Chess is solved.-
smatovic
- Posts: 3599
- Joined: Wed Mar 10, 2010 10:18 pm
- Location: Hamburg, Germany
- Full name: Srdja Matovic
-
Frank Quisinsky
- Posts: 7285
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: SPCC: Classical Ratinglist will be stopped
Hi Stefan,
you are full of power and energy.
New ideas form, come and go.
Most important is that you have fun with your work.
You can be sure I will follow your new activities.
To ratings:
My opinion about it is very old and good known.
Ratings are snapshots!
Unfortunately ... can be manipulated with other opponents and are never exactly that what the most await / expected.
To mix older ratings (engines with older good known ideas) with newer ratings (engines with newer good known ideas) is complicated. Maths alone can't solved this. Furthermore, to work on ratings is a lot of work. New engines and all the updates is a full time job. Electricity is expensive!!
In 45 years computerchess I like most the tourneys. End of a tourney the ratings are history but during a tourney most fun I can generate in watching the games or cheer along with the favorits. Not very important today which engine is on place 1 or 100. Engine on place 100 is stronger as Magnus and no human can beat it. Most important today are the features of the engines, a nice style humans like.
Have thanks for all your work you do for us ... thousands of hours, many years you gave us results and impressions.
That is / was strong because after all ...
Ratings are more or less boring but a "Must have" information for a comptuer chess lover / human brain.
Best
Frank
you are full of power and energy.
New ideas form, come and go.
Most important is that you have fun with your work.
You can be sure I will follow your new activities.
To ratings:
My opinion about it is very old and good known.
Ratings are snapshots!
Unfortunately ... can be manipulated with other opponents and are never exactly that what the most await / expected.
To mix older ratings (engines with older good known ideas) with newer ratings (engines with newer good known ideas) is complicated. Maths alone can't solved this. Furthermore, to work on ratings is a lot of work. New engines and all the updates is a full time job. Electricity is expensive!!
In 45 years computerchess I like most the tourneys. End of a tourney the ratings are history but during a tourney most fun I can generate in watching the games or cheer along with the favorits. Not very important today which engine is on place 1 or 100. Engine on place 100 is stronger as Magnus and no human can beat it. Most important today are the features of the engines, a nice style humans like.
Have thanks for all your work you do for us ... thousands of hours, many years you gave us results and impressions.
That is / was strong because after all ...
Ratings are more or less boring but a "Must have" information for a comptuer chess lover / human brain.
Best
Frank
-
pohl4711
- Posts: 2918
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: SPCC: Classical Ratinglist will be stopped
No, using UHO openings will prevent Computerchess from death. At least for the next decades, IMHO.
And because of this, my testwork will use UHO openings, only, from now. And the best 15 engines are the most interesting engines, so a 15 engine ratinglist is my choice.
Testing will continue on SPCC Computerchess !!!
-
Graham Banks
- Posts: 45482
- Joined: Sun Feb 26, 2006 10:52 am
- Location: Auckland, NZ
Re: SPCC: Classical Ratinglist will be stopped
All the best with your new project. 
gbanksnz at gmail.com
-
pohl4711
- Posts: 2918
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: SPCC: Classical Ratinglist will be stopped
Thanks. But it is no new project, just an expansion of my (already existing) UHO-Top10 Ratinglist to an UHO-Top15 Ratinglist. And the regular testing of Stockfish Dev versions will be continued in this Ratinglist, too.
https://www.sp-cc.de/uho_ratinglist.htm
-
pohl4711
- Posts: 2918
- Joined: Sat Sep 03, 2011 7:25 am
- Location: Berlin, Germany
- Full name: Stefan Pohl
Re: SPCC: Classical Ratinglist will be stopped
I agree. Just one more good reason for me, to change the test-setup: It is not only about using UHO-openings instead of balanced openings, but the new Ratinglist will be a huge RoundRobin-tournament (as my UHO-Top10 Ratinglist already is (https://www.sp-cc.de/uho_ratinglist.htm)):Frank Quisinsky wrote: ↑Tue Sep 26, 2023 9:36 am
To ratings:
My opinion about it is very old and good known.
Ratings are snapshots!
Unfortunately ... can be manipulated with other opponents and are never exactly that what the most await / expected.
16 engines (15 strongest plus latest Stockfish Dev). Each engine will play 1000 games vs. each opponent: 15000 games and each engine has the exact same opponents, because the new ratinglist is a huge RoundRobin-tournament...
And, I will do much more than just printing a ratinglist with some Celos. I will also provide a Gamepair-recalculation of the ratinglist and (of course) I will build an EAS-Ratinglist out of the games with my EAS-tool, so we can learn a lot about the playing-style of the engines and it's weaknesses and strengths.
-
chessica
- Posts: 1062
- Joined: Thu Aug 11, 2022 11:30 pm
- Full name: Esmeralda Pinto
Re: SPCC: Classical Ratinglist will be stopped
I now have a question about the games, what is special about the games? What distinguishes them? Are these special openings?pohl4711 wrote: ↑Tue Sep 26, 2023 8:15 am Important News: I decided to stop doing my SPCC-Ratinglist, because the balanced openings, used for this ratinglist, are not working anymore in these days of superstrong engines... My UHO-Ratinglist will take over here: I will add 5 more engines to this huge RoundRobin-tournament, so a new UHO-Top15 Ratinglist will come in 1 or 2 weeks from now. Because the measurement of Stockfish-Dev progress must be continued, the reference-Elo of this UHO-Top15 Ratinglist will no longer be 0 Elo for Stockfish 16, but 3821 Elo for Stockfish 16 (Elo of SF 16 in my SPCC-Ratinglist).
The EAS-Ratinglist will be calculated out of the UHO-Top15 Ratinglist games in the future.
If you want to download the games of the SPCC-Ratinglist, do it right now, because in 6-10 days, the SPCC-Ratinglist will be removed from my website.
How big is the file?
-
Frank Quisinsky
- Posts: 7285
- Joined: Wed Nov 18, 2009 7:16 pm
- Location: Gutweiler, Germany
- Full name: Frank Quisinsky
Re: SPCC: Classical Ratinglist will be stopped
Hi Stefan,
your idea is strong but I think with 20 engines or better 21 it's more interesting. Many opponents are important also.
What I would like to see later on your site ...
You have 16 engines = 15.000 games per engine from TOP-16.
Now a new rofChade version is out and you put the new version.
End of the day it's interesting to put the older rofChade games in a separate database. So you have for your main list and statistics again and again the 15.000 games per engine. If you mixed over a long time all in one database I think the ratings are not stable enough if you added from different engines (example Stockfish) to many versions.
Biggest problems is to find out again and again which of the engines are in TOP-16. And for your tools it is more interesting to add the engines are stronger in king attacks / combinations. In the current TOP-15 are to many passiv engines (playing style after opening book moves). Here the biggest problems is, to find out again and again which engines are stronger in king attacks. All isn't easy ...
My personal TOP-20 aren't the TOP-20 a rating list produced.
My opinion ... must not be right.
I have start last year the work on a balanced database produced many short games. All my collected short engine games are inside. But I had a hard last year (private things) and forgot the idea a bit. Often I am thinking the idea with unbalanced openings is maybe the better idea in times engines are stronger and stronger. In combination with your idea that all in the list have to play with the same lines with black and white and the same quantity of games.
Wollte mir nicht schon wieder die Rübe runzelig denken und nutzte für mein laufendes Turnier einfach wieder mein altes FEOBOS Buch ohne mich hier wirklich nochmals in Zeug zu legen. Mit mehr Aktivitäten haben wir natürich auch mehr Erfahrung aber letztendlich lernen wir nie aus und werfen oftmals ältere Denkweisen über Board. Geht mir laufend so wenn ich drüber nachdenke, nur bekomme ich dann Kopfschmerzen weil zu viele Stellschrauben ineinander laufen die mir bekannt sind aber es zu zeitaufwendig wird Lösungen zu suchen.
Best
Frank
your idea is strong but I think with 20 engines or better 21 it's more interesting. Many opponents are important also.
What I would like to see later on your site ...
You have 16 engines = 15.000 games per engine from TOP-16.
Now a new rofChade version is out and you put the new version.
End of the day it's interesting to put the older rofChade games in a separate database. So you have for your main list and statistics again and again the 15.000 games per engine. If you mixed over a long time all in one database I think the ratings are not stable enough if you added from different engines (example Stockfish) to many versions.
Biggest problems is to find out again and again which of the engines are in TOP-16. And for your tools it is more interesting to add the engines are stronger in king attacks / combinations. In the current TOP-15 are to many passiv engines (playing style after opening book moves). Here the biggest problems is, to find out again and again which engines are stronger in king attacks. All isn't easy ...
My personal TOP-20 aren't the TOP-20 a rating list produced.
My opinion ... must not be right.
I have start last year the work on a balanced database produced many short games. All my collected short engine games are inside. But I had a hard last year (private things) and forgot the idea a bit. Often I am thinking the idea with unbalanced openings is maybe the better idea in times engines are stronger and stronger. In combination with your idea that all in the list have to play with the same lines with black and white and the same quantity of games.
Wollte mir nicht schon wieder die Rübe runzelig denken und nutzte für mein laufendes Turnier einfach wieder mein altes FEOBOS Buch ohne mich hier wirklich nochmals in Zeug zu legen. Mit mehr Aktivitäten haben wir natürich auch mehr Erfahrung aber letztendlich lernen wir nie aus und werfen oftmals ältere Denkweisen über Board. Geht mir laufend so wenn ich drüber nachdenke, nur bekomme ich dann Kopfschmerzen weil zu viele Stellschrauben ineinander laufen die mir bekannt sind aber es zu zeitaufwendig wird Lösungen zu suchen.
Best
Frank