Computer Chess Club Archives


Search

Terms

Messages

Subject: My thoughts

Author: Axel Schumacher

Date: 11:13:48 01/15/05

Go up one level in this thread


On January 15, 2005 at 08:05:56, Jason Kent wrote:

>I'd like to see axel schumacher give his thoughts on the blitz settings.


O.K. here some thoughts:

Cm10k plays indeed quite different to CM9000, so we have to treat them a little
bit different. I will focus here on CM10k, since it is simply newer, and a
little bit better IMHO.
However, surprisingly, I think the standard settings in CM10k are even weaker
than in previous versions. Nevertheless, I could observe that small changes in
the values had a huge impact on playing strenght. bsasically, the fork is more
pronounced than in CM9000, where it is easier to use extreme values.
So far, the Bltz-setting CM10k Schumacher performs indeed much better than all
other versions. It still has weaknesses. In the moment I let it play at long
time control - and - it still pays very good. Only Shredder is a hard nut to
crack for this settings. This is ostly due to the fact that Shredder more so
than any other engines gains with prolonged time (or it looses with shorter time
controls, everything is relative).
When loosing games I noticed that TheKing engine really looses unusually often
due to passed pawns (especially on the Kingside). So, I would suggest to raise
opponents passes pawn value, even more so for longer time controls, especially
when using TheKing outside of it's own GUI, since it will not be able to access
tablebases. I neglected this setting a bit in the past, though.
I agree with Graham that increasing the pawn weakness too much is not
beneficial. The pawn-play is not CM strenght per se. Furthermore, compared to
other engines, it is a rather slow searcher, so I would rather like to see
TheKing in open positions. In CM10k I would not go over 104 for pw.

Material: I think the material balance with the other setting works better in
CM10k, so I don't advice too much fiddling with these values. Increasing rook
and own queen value somewhat, might help though. Increasing opq is a
disadvantage in my experience.

Still essential for CM10k is to increase the king safety. A value of 150 seemed
to work pretty good.
Also important is the center control. I recommend at least 105, better even 108
or higher, but not too high either (<115). Center control has to be balanced
with mobility. Both values can't be very high at the same time.
Mobility is crucial for CM10k, since it gives opportunities to open up the game.
I recommend at least a 110 better a 120 for mobility.

Selectivity: It has not a big impact on playing strenght in my opinion. I made a
tournament with CM9000 and CM10k with different sel settings and could not find
that the gameplay/strenght changed dramatically. But I may be wrong when it
comes to longer tme-controls, since I test primarily in Blitz. Sel 14 works fine
I think, maybe better than 12 (for CM10k).

avd: I would not change it too much, but again it may help open the position a
little bit by forcing more exchanges. I recommend a value of -2.

Just try :-)

Axel




This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.