Horvath Play Styles
Moderators: Harvey Williamson, Steve B, Watchman
Forum rules
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
This textbox is used to restore diagrams posted with the fen tag before the upgrade.
- spacious_mind
- Senior Member
- Posts: 4001
- Joined: Wed Aug 01, 2007 10:20 pm
- Location: Alabama
- Contact:
- spacious_mind
- Senior Member
- Posts: 4001
- Joined: Wed Aug 01, 2007 10:20 pm
- Location: Alabama
- Contact:
- spacious_mind
- Senior Member
- Posts: 4001
- Joined: Wed Aug 01, 2007 10:20 pm
- Location: Alabama
- Contact:
[fen]rnb1kb1r/3p1pp1/2p2q1p/pp2p3/4P3/1BNP1N2/PPP2PPP/R2QK2R w KQkq - 0 8[/fen]
The Spracklen's really have a problem with this position. Fidelity 68000 Excel and Fidelity 68000 Mach IIb also play 8. Nf3?.
It is only when you get to Fidelity 68000 Excel Mach III that finally the move 8. a3 is played.
I have not tested the Fidelity Mach II L.A. version, but is it possible that only once you get to a Mach III and beyond that this weakness in it's play is fixed?
best regards,
The Spracklen's really have a problem with this position. Fidelity 68000 Excel and Fidelity 68000 Mach IIb also play 8. Nf3?.
It is only when you get to Fidelity 68000 Excel Mach III that finally the move 8. a3 is played.
I have not tested the Fidelity Mach II L.A. version, but is it possible that only once you get to a Mach III and beyond that this weakness in it's play is fixed?
best regards,
Nick
- spacious_mind
- Senior Member
- Posts: 4001
- Joined: Wed Aug 01, 2007 10:20 pm
- Location: Alabama
- Contact:
Hi Steve,Steve B wrote:Hey Nick
what TC are you using ?
i still have the Septennial setup after dispelling that false bug report
perhaps i'll feed it the position
All Dressed Up With No Place To Go Regards
Steve
I am using 30 seconds per move level a3 or a4 depending on fidelity computer.
Thanks
Nick
Nick
- Steve B
- Site Admin
- Posts: 10144
- Joined: Sun Jul 29, 2007 10:02 am
- Location: New York City USofA
- Contact:
OKspacious_mind wrote:Hi Steve,Steve B wrote:Hey Nick
what TC are you using ?
i still have the Septennial setup after dispelling that false bug report
perhaps i'll feed it the position
All Dressed Up With No Place To Go Regards
Steve
I am using 30 seconds per move level a3 or a4 depending on fidelity computer.
Thanks
Nick
at 30 Sec...
Septy considers Nf3 for about 9 Secs then switches to a4 and plays this move
scoring the position ....+.16
September Morn Regards
Steve
- spacious_mind
- Senior Member
- Posts: 4001
- Joined: Wed Aug 01, 2007 10:20 pm
- Location: Alabama
- Contact:
Interesting. I will play CC9, Septennial, Champion, Par Ex and USCF through the tests after I finish with Excel, Mach III and L.A.Steve B wrote:OKspacious_mind wrote:Hi Steve,Steve B wrote:Hey Nick
what TC are you using ?
i still have the Septennial setup after dispelling that false bug report
perhaps i'll feed it the position
All Dressed Up With No Place To Go Regards
Steve
I am using 30 seconds per move level a3 or a4 depending on fidelity computer.
Thanks
Nick
at 30 Sec...
Septy considers Nf3 for about 9 Secs then switches to a4 and plays this move
scoring the position ....+.16
September Morn Regards
Steve
Mach IIb didn't like the test games at all. Scored worse than SU9.
Regards
Nick
- Fernando
- Admiral of the Fleet
- Posts: 3059
- Joined: Tue Jul 31, 2007 4:35 pm
- Location: Santiago de Chile
False report?Steve B wrote:Hey Nick
what TC are you using ?
i still have the Septennial setup after dispelling that false bug report
perhaps i'll feed it the position
All Dressed Up With No Place To Go Regards
Steve
perhaps bad memory....
You insist in considering me a lier.
There is not enough money in the world to pay me the compensation I will ask in NY court regards
Fern
Festina Lente
Hi,spacious_mind wrote:Hi Paul,
Well it sounds like you are having fun trying out the styles. It is not so easy understanding them.
I recently bought a Systema Challenge and wanted to tune the evaluation. Not to squeeze more strenght, but to slightly adjust the playing style towards a more active one. Indeed the "Challenge" is to understand the effect of the settings!
For comfort, I "refresh" here the available style settings Nick shared top of the thread:
First, a comment on setting all the features to same value but different to the default one (value 10 regarding the Challenge): pieces material values are not part of the settable features, but for sure are part of the evaluation (some other measures as well). So shifting the settable values have a weighting effect with regards to non settable ones. And I guess the range 0-99 is non linear, allowing to raise or lower the 10 default value by a 10 ratio + option to disable it (0). So relative weight 10 can be divided by 10 (new weight=1) or multiplied by 10 (close to, new weight=99).spacious_mind wrote:Now back to the 9 standard built-in style settings. If you change the style on a Horvath and then go and look inside the configurable style option, you will see the following parameter settings:
The standard out of the box parameter is style 5 which everyone plays with. If you notice the settings for style 5, every parameter is set to the value 10.
I used the Ketterling's initiativ test with level 7 (2s) as short time control helps avoiding tactical influence on evaluation (and a H8 is a fast one).
Non-described parameters use default setting.
Some obvious symetries exist in the results I got (style 1-9, F3 0-99...).
Style definitions (as per user's manual) do not appear to be obviously consistent with results (most "aggressive" side according to style setting enjoys drawing). And when the program is up in material, it knows how to avoid a draw (by the way, I stopped any game at max 60 moves, so draw in 60 is sort of a high-value meaning 60 or more).
I also tested Paul's idea (all features set to 20). Not a bad one... But I chased more understanding, and smallest change with large effect.
I decided rather quickly not to change F3 (king attack) as it has little positive impact on initiative; when shifted away from default value, it provides very unbalanced results depending on the color the program plays.
Then I focused on the promising F4/F5 settings (aggressive play by White's/Black's pieces). Promising but tricky, as these are set for each color, while other settings fit both. To make it short, a twice reduced values there (5 instead of 10) provided results that fit my taste (not only the number of moves to mate, but alse the played moves). In addition, lowering the value seems consistent with styles definition. Increasing it led to draws.
Final step I made was checking the playing level, using Nick's 5 games test.
I used level 20 (fixed 15s). Little changes occurred in the chosen moves, leading to roughly same performance. I expected not to hurt the playing level, seems fine.
Should you be interested with, games are available here.
Kind regards,
Tibono
- mclane
- Senior Member
- Posts: 1605
- Joined: Sun Jul 29, 2007 9:04 am
- Location: Luenen, germany, US of europe
- Contact:
The playing styles of the Horvath program are very complex, aren’t they ?
The job is IMO to increase the activity of the defensive and positional playing program the way it gets not overrun each game.
This is difficult because in the moment you increase activity it begins to use the queen and puts it to b6 and runs arround with it,
I tried Styles in my past
Long time ago,
And I have a style of my friend uli in a tournament , but not very successful so far,
I am open for any new idea,
The job is IMO to increase the activity of the defensive and positional playing program the way it gets not overrun each game.
This is difficult because in the moment you increase activity it begins to use the queen and puts it to b6 and runs arround with it,
I tried Styles in my past
Long time ago,
And I have a style of my friend uli in a tournament , but not very successful so far,
I am open for any new idea,
What seems like a fairy tale today may be reality tomorrow.
Here we have a fairy tale of the day after tomorrow....
Here we have a fairy tale of the day after tomorrow....
- paulwise3
- Senior Member
- Posts: 1508
- Joined: Tue Jan 06, 2015 10:56 am
- Location: Eindhoven, Netherlands
Hi Tibono,
Thanx for your post! I did try testing with some feature settings back from 20 to 10, but didn't take the time for extensive testing. Setting f4/f5 to less then 10 is an interesting idea. Maybe it could lead to less focus on material gain.
Also I think it could be important not to set them both at the same value, but choose values depending on with which color it plays.
I might get inspired to do more testing this year...
Best regards,
Thanx for your post! I did try testing with some feature settings back from 20 to 10, but didn't take the time for extensive testing. Setting f4/f5 to less then 10 is an interesting idea. Maybe it could lead to less focus on material gain.
Also I think it could be important not to set them both at the same value, but choose values depending on with which color it plays.
I might get inspired to do more testing this year...
Best regards,
2024 Special thread: viewtopic.php?f=3&t=12741
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
Hi, I achieved an interesting setting (to my humble opinion!); you can read about it on my webpage here.
Have fun,
Tibono
Have fun,
Tibono
- paulwise3
- Senior Member
- Posts: 1508
- Joined: Tue Jan 06, 2015 10:56 am
- Location: Eindhoven, Netherlands
Hi Tibono,
Thank you, that looks very interesting and promising!
I will do some tests with these settings too in the coming weeks. Will let you know the results.
Horvath regards,
Paul
Thank you, that looks very interesting and promising!
I will do some tests with these settings too in the coming weeks. Will let you know the results.
Horvath regards,
Paul
2024 Special thread: viewtopic.php?f=3&t=12741
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...
2024 Special results and standings: https://schaakcomputers.nl/paul_w/Tourn ... 25_06.html
If I am mistaken, it must be caused by a horizon effect...