Hamsters CCRL rating

[Uri Blass]

I looked at the latest ccrl rating
http://computerchess.org.uk/ccrl/4040.l ... t_all.html

Hamsters 0.4 2707 +136 −115 79.6% −220.8 18.5% 27


For comparison rating at 40/4 from
http://computerchess.org.uk/ccrl/404/ra ... t_all.html
50 Hamsters 0.4 2569 +29 −29 41.6% +60.7 28.4% 412
55.8%

I wonder if it is only luck or maybe hamsters performs relatively
better at long time control.
I know that hamsters play against relatively weak opponents but you need to win against them in order to achieve high rating and it is not trivial.

Hamster clearly does this job better than programs with better blitz rating like
Twisted Logic 0.099x or BugChess2 1.5.1 in the following tournament:
http://kirr.homeunix.org/chess/discussi ... a&start=25

I do not believe that the rating of hamsters will remain that high but
my guess based on the results that I read is that hamsters like old movei(00_8_295) does relatively better at long time control

In the blitz rating list we have
Movei 0.08.295 2571 +25 −25 42.1% +54.5 29.7% 539
Hamsters 0.4 2569 +29 −29 41.6% +60.7 28.4% 412

In the long time control rating list we have
Movei 0.08.295 2621 +37 −37 41.5% +59.2 32.0% 247
53.5%

What is your prediction?
Do you expect hamsters to get long time control that is higher than
movei00.8.295 after enough games?

Uri

[rdan1987]

I looked at the latest ccrl rating
http://computerchess.org.uk/ccrl/4040.l ... t_all.html

Hamsters 0.4 2707 +136 −115 79.6% −220.8 18.5% 27


For comparison rating at 40/4 from
http://computerchess.org.uk/ccrl/404/ra ... t_all.html
50 Hamsters 0.4 2569 +29 −29 41.6% +60.7 28.4% 412
55.8%

I wonder if it is only luck or maybe hamsters performs relatively
better at long time control.

How can you get conclusions after only 27 games??

Don't you know that CCRL needs months for an engine to get more than 300 games??.....Probably in 2008 Hamsters 0.4 will have enough games so you can have proper conclusions!!!

[Spock]

I looked at the latest ccrl rating
http://computerchess.org.uk/ccrl/4040.l ... t_all.html

Hamsters 0.4 2707 +136 −115 79.6% −220.8 18.5% 27


For comparison rating at 40/4 from
http://computerchess.org.uk/ccrl/404/ra ... t_all.html
50 Hamsters 0.4 2569 +29 −29 41.6% +60.7 28.4% 412
55.8%

I wonder if it is only luck or maybe hamsters performs relatively
better at long time control.

How can you get conclusions after only 27 games??

Don't you know that CCRL needs months for an engine to get more than 300 games??.....Probably in 2008 Hamsters 0.4 will have enough games so you can have proper conclusions!!!

Fortunately, it has 900 games on the CCRL FRC list
http://computerchess.org.uk/ccrl/404FRC/

but that's blitz as well......


.

[Uri Blass]

I looked at the latest ccrl rating
http://computerchess.org.uk/ccrl/4040.l ... t_all.html

Hamsters 0.4 2707 +136 −115 79.6% −220.8 18.5% 27


For comparison rating at 40/4 from
http://computerchess.org.uk/ccrl/404/ra ... t_all.html
50 Hamsters 0.4 2569 +29 −29 41.6% +60.7 28.4% 412
55.8%

I wonder if it is only luck or maybe hamsters performs relatively
better at long time control.

How can you get conclusions after only 27 games??

Don't you know that CCRL needs months for an engine to get more than 300 games??.....Probably in 2008 Hamsters 0.4 will have enough games so you can have proper conclusions!!!

I did not say that I am sure about something but I can give an opinion.

Of course number of games is not enough to be sure but it is enough for me to get the opinion that hamsters perform better at long time control.

I can change my mind if I see more data that suggest different things.

Uri

[Vempele]

How can you get conclusions after only 27 games??

Don't you know that CCRL needs months for an engine to get more than 300 games??.....Probably in 2008 Hamsters 0.4 will have enough games so you can have proper conclusions!!!

Look at the error bars. Statistically Hamsters 0.4 is extremely likely to be stronger at 40/40. Besides,

I wonder if it is only luck or maybe hamsters performs relatively
better at long time control.

doesn't look like a conclusion anyway.

[Dirt]

Look at the error bars. Statistically Hamsters 0.4 is extremely likely to be stronger at 40/40.

You don't need to just look at the error bars. Using the CCRL custom comparison utility the Likelihood of Superiority of Hamsters at 40/40 is only 78.5%.

[Uri Blass]

Look at the error bars. Statistically Hamsters 0.4 is extremely likely to be stronger at 40/40.

You don't need to just look at the error bars. Using the CCRL custom comparison utility the Likelihood of Superiority of Hamsters at 40/40 is only 78.5%.

I wonder how did you get the 78.5% number.
I do not know about CCRL utility that compare rating at different time control.

I only know about liklihood of superiority between 2 programs from the same list and even in this case I know to do it only when they are from the same family and even for programs for the same family I cannot always do it and I cannot see lilkelihood of superiority between hamsters and kiwi.

Uri

[Dirt]

I noticed after I posted, but too late to edit, that there are four more games in the list you linked to, which will make some difference. The 78.5% came from here: http://computerchess.org.uk/ccrl/4040/c ... =by+rating

I hope that link works for other people.

[Alessandro Scotti]

There are not enough games. Last time I tested, Hamsters performed slightly worse at longer time controls, but that was before a bunch of many bug fixes. I expect the rating to come close to that of blitz, hopefully not worse this time.

The good news of course is that whatever the rating, it made it into the CCRL list!

[Graham Banks]

There are not enough games. Last time I tested, Hamsters performed slightly worse at longer time controls, but that was before a bunch of many bug fixes. I expect the rating to come close to that of blitz, hopefully not worse this time.

The good news of course is that whatever the rating, it made it into the CCRL list!

Hamsters 0.4 will steadily get more games and reasonably quickly.

Another tournament starts this weekend (Rising Stars v Fading Warriors 2007) for example.
Next week, a further tournament (Bright Revelations) will start in which Hamsters is also included.
Those are just from my end as one tester.

I'm looking forward to seeing how your latest creation goes Alessandro. Keep up the great work!

Regards, Graham.