It does not change anything in the EAS score, so I assume you don't use the Plycount tag as the number of moves in a game.
OTOH the number of shorties now is a lot bigger than your tool which checks for EAS points till move 50 and creates 1325 shorties while my tool checks till move 40 and gets 2573 shorties, see above.
matejst and Ghppn like this post
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sat Jul 08, 2023 9:50 am
UPDATE
Cut the PGN of won games at the point the game is decided.
------------------------------------------------------------------- *** Average length of all won games: 68 moves *** Engine gets bonuspoints, if its avg. won games length is shorter *** Engine gets maluspoints, if its avg. won games length is longer
*** Average length of all won games: 47 moves *** Engine gets bonuspoints, if its avg. won games length is shorter *** Engine gets maluspoints, if its avg. won games length is longer
Ghppn likes this post
pohl4711
Posts : 160 Join date : 2022-03-01 Location : Berlin
Subject: Re: EAS - for Stefan Sat Jul 08, 2023 1:49 pm
Admin wrote:
Wrote a tool that calculates at which move a game is decided and stores the result as 'ply' in the PGN tag 'Plycount'. From a 16.000 game PGN:
It does not change anything in the EAS score, so I assume you don't use the Plycount tag as the number of moves in a game.
OTOH the number of shorties now is a lot bigger than your tool which checks for EAS points till move 50 and creates 1325 shorties while my tool checks till move 40 and gets 2573 shorties, see above.
The EAS-Tool uses the summary-tool by Norm Pollock to find the average length of games in a pgn. And mention, evals of engines are a nice thing, but each engine has different evals, so this is not a valid number/data for doing anything in the EAS-Tool. And the format of eval in the pgn-comments can differ, using different GUIs... Mention, the EAS-tool is not only for engine-games (pgn, played in any GUI), but also for human games. Because of this, the EAS-Tool operates only on the moves and the result of games in a pgn. Nothing else. No evals, no comments, no tags- all this is ignored (except the result-tag and the player-names tags (white/black)) . And not a single game in a pgn is cutted or manipulated in any way (except deleting all comments (but this is only for faster computing)). And it is important, to do so, if you want a tool, that can operate on any chess game (by humans or engines (played in any GUI)).
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 10:26 am
pohl4711 wrote:
Admin wrote:
Wrote a tool that calculates at which move a game is decided and stores the result as 'ply' in the PGN tag 'Plycount'. From a 16.000 game PGN:
It does not change anything in the EAS score, so I assume you don't use the Plycount tag as the number of moves in a game.
OTOH the number of shorties now is a lot bigger than your tool which checks for EAS points till move 50 and creates 1325 shorties while my tool checks till move 40 and gets 2573 shorties, see above.
The EAS-Tool uses the summary-tool by Norm Pollock to find the average length of games in a pgn. And mention, evals of engines are a nice thing, but each engine has different evals, so this is not a valid number/data for doing anything in the EAS-Tool. And the format of eval in the pgn-comments can differ, using different GUIs... Mention, the EAS-tool is not only for engine-games (pgn, played in any GUI), but also for human games. Because of this, the EAS-Tool operates only on the moves and the result of games in a pgn. Nothing else. No evals, no comments, no tags- all this is ignored (except the result-tag and the player-names tags (white/black)) . And not a single game in a pgn is cutted or manipulated in any way (except deleting all comments (but this is only for faster computing)). And it is important, to do so, if you want a tool, that can operate on any chess game (by humans or engines (played in any GUI)).
1. Regarding the violet, I wrote my own PGN parser in the mid 90, it supports all GUI output, so your assumption is wrong.
2. The formula that decides the move number a game is won is pretty rock solid, so far I haven't seen a false case. But you might draw your own conclusion when I am done and will publish the tool.
3. Another advantage of score evaluation, you can skip won games from bad openings line, your tool can't do that.
Look, I am not criticizing your great tool, I am giving constructive feedback, it's up you if you want to consider it.
In the next post an example from your database.
Mclane likes this post
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 10:44 am
Made a selection from your database, 68.000 games, top engines only.
Left the original list, right the updated list and note the reshuffled ranking effect based on the much lowered average wins, compare the 2 average win columns.
pohl4711
Posts : 160 Join date : 2022-03-01 Location : Berlin
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 12:48 pm
So what? Different result doesnt mean better result. I play my ratinglist games without adjucation for good reasons. Adjucate the games after they are finished would be crazy, IMO. And I believe, a real aggressive playing engine must not only reach won positions fast, but should win and mate fast after doing so, too. So, this should be measured. And thats what I am doing.
The EAS Tool works fine for me as it is. And I will not change it or my EAS Ratinglist or ending the games in my testruns by adjucation.
But you can do it, if you want. You are a much better programmer, than I am. And the code of EAS Tool is not complicated and I commented each operation. So, it should be easy for you, to change the code in the Batchfile. So, feel free to make your own EAS Tool version. Fine for me.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 5:11 pm
Admin wrote:
Another advantage of score evaluation, you can skip won games from bad openings line
I've been doing that for 10 years, but without a tool that can process a clean input PGN and output a PGN where the games end at the first mistake, I've had to do it in a partially manual way. Quite a hassle, as you can imagine.
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 7:35 pm
Ozymandias wrote:
Admin wrote:
Another advantage of score evaluation, you can skip won games from bad openings line
I've been doing that for 10 years, but without a tool that can process a clean input PGN and output a PGN where the games end at the first mistake, I've had to do it in a partially manual way. Quite a hassle, as you can imagine.
And you never found a tool that did the work for you?
Last edited by Admin on Sat Jul 15, 2023 9:55 am; edited 1 time in total
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 8:20 pm
Maybe I wasn't looking for it in the right place.
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 8:36 pm
In case you have not, SOMU can do the job for you.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 8:59 pm
I doesn't look like this would take a clean (no comments) PGN, analyze the games and save them up to the ply previous to where the first mistake was made.
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 10:33 pm
Maybe the POLY util is your friend, using the [F5] option Check for bad book openings
It checks the first 5 moves (defined by Depth) after the book opening.
But... it needs comments.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sun Jul 09, 2023 10:46 pm
That's usually the problem. I always end resorting to full game analysis, which is cumbersome and also inefficient, as it will keep analyzing after the mistake.
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Mon Jul 10, 2023 8:21 am
Tell me how the output should look like and I will look if I have it somewhere, else I can make it.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Mon Jul 10, 2023 11:32 am
There are more mistakes in the remaining moves, but analysis should stop as soon as Qxg6 is identified as a mistake, and the game stored in the previous position.
Engine parameters should be left at default values.
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Tue Jul 11, 2023 10:15 am
Subject: Re: EAS - for Stefan Sat Jul 15, 2023 10:30 am
Will do. Any particular reason for going with SF's previous version?
Admin Admin
Posts : 2608 Join date : 2020-11-17 Location : Netherlands
Subject: Re: EAS - for Stefan Sat Jul 15, 2023 10:40 am
Ozymandias wrote:
Will do. Any particular reason for going with SF's previous version? Can it be substituted by SF16?
Just copy sf16.exe into the somu folder and use the option (E)ngine to load SF16 or whatever SF version.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sat Jul 15, 2023 10:52 am
Already noticed, did and edited my previous post. ;-)
I introduced a low number for the ms/depth parameter and was identified as depth. 100 and up is ms?
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sat Jul 15, 2023 11:10 am
Using a PGN with a few games, I see a message "Games: 10" as soon as it's selected, even thought the PGN has over a 100. Also, because there were no blunders in the first game, it doesn't save a truncated PGN and can't proceed. If the requirements are lowered, it saves it but doesn't analyze the subsequent games.
Ozymandias
Posts : 622 Join date : 2020-11-23
Subject: Re: EAS - for Stefan Sat Jul 15, 2023 11:14 am
Forgot to mention that it analyses the game until the end, independently of when/if a blunder is found.
Home 
