Search found 302 matches

by UncombedCoconut
Sat Jan 30, 2010 5:11 pm
Forum: Computer Chess Club: General Topics
Topic: Timecontrol, Win- & DrawRatio
Replies: 6
Views: 443

Re: Timecontrol, Win- & DrawRatio

But then I also compared the win ratio for white and found out that it was staying equal or even increasing slowly. I would have assumed they would decrease and tend to reach 50% as draw rate continues to rise, but this is the contrary. The data would suggest that the likelihood for white to win in...
by UncombedCoconut
Sat Jan 30, 2010 12:51 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: New Year's resolution: a chess engine written in BASH script
Replies: 67
Views: 34843

Re: New Year's resolution: a chess engine written in BASH sc

It's stoopid, I know. But, it's been done in Javascript. I would like to see an obviously very basic script that searches 3 or 4 ply, with simple piece value eval. A minimax and no alpha-beta, etc. Anybody out there who could help? When my attention drifts from coaxing Brutus to compile natively on...
by UncombedCoconut
Thu Jan 28, 2010 11:21 pm
Forum: Computer Chess Club: General Topics
Topic: Brutus Updated
Replies: 14
Views: 2518

Re: Brutus Updated

I can give this a try, although I'm sure others are better qualified. I like open-source engines. :)
by UncombedCoconut
Wed Jan 27, 2010 11:06 pm
Forum: Computer Chess Club: General Topics
Topic: Unstoppable vs Untouchable pawns in a pawnrace?
Replies: 4
Views: 594

Re: Unstoppable vs Untouchable pawns in a pawnrace?

zullil wrote:Stockfish-1.6.2 (8 threads) seemed to agree with you, just before a stray cosmic ray killed it.

Code: Select all

info nodes 3809368776 nps 10099336 time 377190 hashfull 999
Segmentation fault
Irreproducible bugs are always the most fun... I've never gotten SF to do that!
by UncombedCoconut
Mon Jan 18, 2010 12:34 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine Testing - Statistics
Replies: 55
Views: 7882

Re: Engine Testing - Statistics

I was taking conservative values: eg LOS when 20 points ahead after 200 games is 95.67% or 95.50% when 14 points ahead after 100 games. So maybe if you combine the two you actually don't need to be 34 points ahead, but 33 would do. But that is no major problem I think. Or what other false positives...
by UncombedCoconut
Sun Jan 17, 2010 11:23 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine Testing - Statistics
Replies: 55
Views: 7882

Re: Engine Testing - Statistics

You say that there is no way, and that's false, and very easy to prove. Here is counter-example: Your simpler example is nice. :D BTW, a literal sequential probability ratio test is unsuitable for testing "engine A is stronger than engine B" vs "engine A is not stronger than engine B". It requires ...
by UncombedCoconut
Sun Jan 17, 2010 11:10 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine Testing - Statistics
Replies: 55
Views: 7882

Re: Engine Testing - Statistics

I'll demonstrate it again with the figures form my LOS table; alpha = 5%; tournament size 200 games (predefined); question whether to stop after 100 games: According to the table you need to be 20 games ahead after 200 games to be superiour regarding the 5% margin. If we now still have 100 games to...
by UncombedCoconut
Sun Jan 17, 2010 10:43 pm
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine Testing - Statistics
Replies: 55
Views: 7882

Re: Engine Testing - Statistics

You are doing the same thing with the Elo information. It goes up and down, as the data I posted clearly shows. If you stop when it looks good, the test is meaningless. yes, if you totally bust the code it will plummet like a rock and you could stop quickly when the results look horrible. But to ma...
by UncombedCoconut
Sun Jan 17, 2010 11:53 am
Forum: Computer Chess Club: Programming and Technical Discussions
Topic: Engine Testing - Statistics
Replies: 55
Views: 7882

Re: Engine Testing - Statistics

Maybe I am doing something wrong here, but I am still not getting quite the same values. eg n=100; alpha=5% --> should be 57.5% I guess it is a bit complicated. To match the table's 5% column I translate as follows: First, since that page discusses 1-sided tests and my program wants to do 2-sided t...
by UncombedCoconut
Sun Jan 17, 2010 12:58 am
Forum: Computer Chess Club: General Topics
Topic: My views on moderation
Replies: 41
Views: 4747

Re: My views on moderation

My views on moderation: Promoter of a commercial (Chess) software should not moderate here! Thank you. Correct ! I second that....dogs of commercial authors should be banned too.... R ight O r L eft F lipside regards, Dr.D :wink: What happened to this : Thanks Graham and you're right....I stop this...