Tool for tactical improvement

[arturo100]

Before hacking it by myself, I would like to know if there is a program that goes through a pgn file and makes an epd file of the positions in which a tactical shot was missed.

I prefer open source software in linux environment, but other solutions could be OK. The pipeline could be something like the following:

1) Feed to crafty annotate function the pgn file.
2) Post-process the crafty log file in order to capture those positions that contains missed tactical shots (parametrized value).
3) Create the epd file of these position.

User could than view the generated epd file in databases like scid that allow to mask the best move and try to guess it.

Thanks in advance.

Arturo

[metax]

I don't know any program like that, but I would greatly appreciate if you would develop one and release it

[Mincho Georgiev]

If I remember correctly, Jim Ablett or Dann Corbit had some time ago something similar in their pgn tool collections. I'm sorry for that incomplete information, but that's all I can remember.

[Guenther]

Before hacking it by myself, I would like to know if there is a program that goes through a pgn file and makes an epd file of the positions in which a tactical shot was missed.

I prefer open source software in linux environment, but other solutions could be OK. The pipeline could be something like the following:

1) Feed to crafty annotate function the pgn file.
2) Post-process the crafty log file in order to capture those positions that contains missed tactical shots (parametrized value).
3) Create the epd file of these position.

User could than view the generated epd file in databases like scid that allow to mask the best move and try to guess it.

Thanks in advance.

Arturo

How should the program recognize if a 'tactical shot' was missed
or there simply was a blunder?
Anyhow there is a program which recognizes jumps in the eval and
you can yourself adjust by how much and for how much moves.
It is possible that you find that way what you wanted. BTW it also
measures the difference of evaluation between two programs,
which could be also helpful for your task.

Get it here:

Thomas McBurneys GameAnalyzer:
(Note that the info that you must extract the pgns from WB debug files
is outdated since the end of 2006 - meanwhile it should take most
standard pgn files as input, not sure though for CBs strange 'new'
%% output pgn files, the old CB pgn output should be ok)
http://home.pacific.net.au/~tommyinoz/gameanalyser.html

If you have questions about that tool, feel free to ask.

Guenther

[arturo100]

Before hacking it by myself, I would like to know if there is a program that goes through a pgn file and makes an epd file of the positions in which a tactical shot was missed.

I prefer open source software in linux environment, but other solutions could be OK. The pipeline could be something like the following:

1) Feed to crafty annotate function the pgn file.
2) Post-process the crafty log file in order to capture those positions that contains missed tactical shots (parametrized value).
3) Create the epd file of these position.

User could than view the generated epd file in databases like scid that allow to mask the best move and try to guess it.

Thanks in advance.

Arturo

How should the program recognize if a 'tactical shot' was missed
or there simply was a blunder?
Anyhow there is a program which recognizes jumps in the eval and
you can yourself adjust by how much and for how much moves.
It is possible that you find that way what you wanted. BTW it also
measures the difference of evaluation between two programs,
which could be also helpful for your task.

Get it here:

Thomas McBurneys GameAnalyzer:
(Note that the info that you must extract the pgns from WB debug files
is outdated since the end of 2006 - meanwhile it should take most
standard pgn files as input, not sure though for CBs strange 'new'
%% output pgn files, the old CB pgn output should be ok)
http://home.pacific.net.au/~tommyinoz/gameanalyser.html

If you have questions about that tool, feel free to ask.

Guenther

Thanks, Guenther. This tool will do the job. Tomorrow, I'll give it a try and I'll let you know if it behaves.
I am interested in catching blunders as well, so just by setting a +2 pawn jump will be sufficient.
The reason why I was was looking for this program is that I noticed that playing against a severely crippled crafty version, I missed several good moves that are actually quickly found by other programs on faster hardware. So, I want to let analyze my games by these programs and be presented again with these same positions. If still I cannot solve them, I am really hopeless!
Thanks again.
Arturo

[arturo100]

Before hacking it by myself, I would like to know if there is a program that goes through a pgn file and makes an epd file of the positions in which a tactical shot was missed.

I prefer open source software in linux environment, but other solutions could be OK. The pipeline could be something like the following:

1) Feed to crafty annotate function the pgn file.
2) Post-process the crafty log file in order to capture those positions that contains missed tactical shots (parametrized value).
3) Create the epd file of these position.

User could than view the generated epd file in databases like scid that allow to mask the best move and try to guess it.

Thanks in advance.

Arturo

How should the program recognize if a 'tactical shot' was missed
or there simply was a blunder?
Anyhow there is a program which recognizes jumps in the eval and
you can yourself adjust by how much and for how much moves.
It is possible that you find that way what you wanted. BTW it also
measures the difference of evaluation between two programs,
which could be also helpful for your task.

Get it here:

Thomas McBurneys GameAnalyzer:
(Note that the info that you must extract the pgns from WB debug files
is outdated since the end of 2006 - meanwhile it should take most
standard pgn files as input, not sure though for CBs strange 'new'
%% output pgn files, the old CB pgn output should be ok)
http://home.pacific.net.au/~tommyinoz/gameanalyser.html

If you have questions about that tool, feel free to ask.

Guenther

Thanks, Guenther. This tool will do the job. Tomorrow, I'll give it a try and I'll let you know if it behaves.
I am interested in catching blunders as well, so just by setting a +2 pawn jump will be sufficient.
The reason why I was was looking for this program is that I noticed that playing against a severely crippled crafty version, I missed several good moves that are actually quickly. found by other programs on faster hardware. So, I want to let analyze my games by these programs and be presented again with these same positions. If still I cannot solve them, I am really hopeless!
Thanks again.
Arturo

Oh, oh... I spoke too early. LGDEBPGN.exe program is no longer available along with Lyapko George's whole web site. Any workaround ?
Thanks again.
Arturo

[David Dahlem]

Oh, oh... I spoke too early. LGDEBPGN.exe program is no longer available along with Lyapko George's whole web site. Any workaround ?
Thanks again.
Arturo

I've done some preliminary testing of GameAnalyser, and it works with Arena games that have the engines evaluation scores included. So LGDEBPGN.exe would not be needed for these games.

[Guenther]

The LGDEBPGN tool is not necessary anymore as I wrote before.
(It is obsolote since end of 2006, its use was to produce readable pgns out of WB debugs, thus you don't need it. Just feed your pgn file and fiddle with
the settings.)

Guenther

[Guenther]

The LGDEBPGN tool is not necessary anymore as I wrote before.
(It is obsolote since end of 2006, its use was to produce readable pgns out of WB debugs, thus you don't need it. Just feed your pgn file and fiddle with
the settings.)

Guenther

Perhaps a caveat, I see you want to check your own games. In this case
of course you only have the evals of one program, I hope this will be
sufficient for your task :)

Guenther

[Norm Pollock]

The "Blunder Finder" of "Game Analyser" is an amazing program. What it does is list "blunder candidates". It is still up to the user to visually check each "blunder candidate" move and make the final decision as to whether or not the "blunder candidate" is really a blunder. For example, when the evaluating engine suddenly sees the checkmate coming in a clearly losing position, it will drastically lower the evaluation and this might give the appearance of a "blunder candidate". But since the game was clearly lost anyway, that move is not a blunder. In other words, it's a "false positive".

An interesting aspect of this program is that it detects "blunder candidates" that were not seen and/or were not taken advantage of by the opposing player.

The weakness of the program is that the user sets an arbitrary bar for what is a "blunder candidate". This will allow many moves just below the bar to go undetected. For example, if I set 750 millipawns as the bar for a "blunder candidate", then a careless loss of a Rook might not be classified as a "blunder candidate" since it will usually only cause a downward evaluation of 500 millipawns. But if you lower the bar, then you start getting more "false positives".