Arimaa Forum
        (http://arimaa.com/arimaa/forum/cgi/YaBB.cgi)
      

        Arimaa >> General Discussion >> Thoughts on my standard set up
        
(Message started by: Kraizy_Dave on Aug 24th, 2014, 8:21am)
Title: Thoughts on my standard set up
Post by Kraizy_Dave on Aug 24th, 2014, 8:21am
Over the last few years my favourite set up has been the following:

R H C M E C H R
R D R R R R D R

The key point for me is that I can attack with my elephant and horse on either flank and as soon as the horse advances a dog can hold b3 or g3. The rabbits are useful on the flanks to free any of my pieces (typically a horse or camel) that are dragged forward to squares that I don't like. The central camel can easily switch flanks depending on how the game develops in the early turns.

Has anyone else tried this set up? Do you like it? I like it as both colours but I suspect that it is strongest as gold.
Title: Re: Thoughts on my standard set up
Post by Fritzlein on Aug 24th, 2014, 10:32am
In the old days there were no EH attacks, only lone-elephant attacks which generally could only achieve rabbit pulls.  In that environment, it was not a good idea to have central rabbits on d1 and e1, because the opposing elephant could pull them out without decentralizing, unlike the a2 and h2 rabbits which could be pulled but required the elephant to leave the middle.  As a consequence, back central rabbits essentially disappeared from setups except when all rabbits were kept on the back row.

Nowadays even when there are rabbit pulls in the opening, it is often a flank horse doing the job rather than the elephant, so the disadvantage of back central rabbits seems negligible, if it is real at all.  If the dogs are preferred on the flanks rather than in the center, then b1 and g1 are better squares than d1 and e1.

I am curious to see whether the bias against back central rabbits will erode further over time, or whether there will continue to be a preference for dogs and/or cats on d1 and e1.
Title: Re: Thoughts on my standard set up
Post by Kraizy_Dave on Aug 24th, 2014, 11:14am
It is that change in a fashion from a defensive style to a more aggressive style that is why I find it surprising that more people don't try this setup. With dogs on d1 and e1 (as appears to be the most common setup) then it only leaves very weak pieces on b3 or g3, whichever side the horse advances on. With an all rabbit back row the dog can get to b3 or g3 just as quickly but the two rabbits are that little bit further back. I think that as rabbits usually advance down the flank then I would prefer them on a1 and a2 to a1 and b1. I think it is likely to save one useful step at some point in the game (e.g. if the rabbit is needed to free a piece on b4 or b5 then whether it takes two or three turns to get to a4 or a5 can make a big difference).
Title: Re: Thoughts on my standard set up
Post by Kraizy_Dave on Aug 24th, 2014, 11:49am
On a loosely linked note, does anyone have a script that could work out what is the most successful setup for gold based on the following criteria:
1. Only count games between two humans
2. Only count setups that have been used by 3 people or more
3. Don't work out the total winning percentage across all games. Instead, work out the winning percentage for each player and then take the average

The idea of the last point is that if one or two very strong payers like certain setups then I don't want to completely dictate the setups score because they probably would have won most of their games with any setup.
Title: Re: Thoughts on my standard set up
Post by Manuel on Aug 25th, 2014, 7:04am
My default setup is more or less the same for more or less the same reasons (I like to bring my flank rabbits a bit forward actively, not only to bring other pieces back, but to actively take part in the attacks and be available for goal threats.
However, I noticed after some time that after my elephant and camel have left for action, I would experience a lack of non-rabbit pieces in the middle, so nowadays I put one cat behind my elephant:

R H C M E R H R
R D R R C R D R
Title: Re: Thoughts on my standard set up
Post by chessandgo on Aug 27th, 2014, 2:13pm
I used to play this this (maybe in 2008?), and also tried half that (one Dog on e1/d1 and one on b1/g1). Over the years I've come to like advancing Dogs in the center in the middle game though, and this setup is a lot less flexible with no piece in the center. A Dog protects c3 just as well from d3 as from b3, and it's kind of rare that you'd want to charge up the b-file with a Horse AND then with a Dog.

So I'd say this setup is very close in strength to standard 99of9, but less flexible.
Title: Re: Thoughts on my standard set up
Post by aaaa on Aug 27th, 2014, 5:00pm

on 08/24/14 at 11:49:50, Kraizy_Dave wrote:
On a loosely linked note, does anyone have a script that could work out what is the most successful setup for gold based on the following criteria:
1. Only count games between two humans
2. Only count setups that have been used by 3 people or more
3. Don't work out the total winning percentage across all games. Instead, work out the winning percentage for each player and then take the average

The idea of the last point is that if one or two very strong payers like certain setups then I don't want to completely dictate the setups score because they probably would have won most of their games with any setup.

I followed your criteria (with a few sensible modifications) and I got four setups with perfect scores, none being used more than 6 times:

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMRDDEHH/RCRRRCRR
3 players, 6 games

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHDMEHDR/RRRCCRRR
5 players, 5 games

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMCHECHR/RDRRDRRR
3 players, 3 games

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMDHDEHR/RRRCCRRR
3 players, 3 games
Title: Re: Thoughts on my standard set up
Post by Kraizy_Dave on Aug 28th, 2014, 1:07pm
Cool. Thanks for sharing that.

Now we finally have proof that swapping a dog and a horse in the 99of9 setup is invincible! It is incredible that so few people have attempted to make that very small change to a common setup.

I must admit that I was hoping that the top results would be non perfect scores. What happens if the minimum number of games and/or the minimum number of players is higher, e.g. 20?
Title: Re: Thoughts on my standard set up
Post by aaaa on Aug 28th, 2014, 4:01pm
Here are all positive-scoring setups that have no strict betters in terms of the three parameters:

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHDMEHDR/RRRCCRRR
5 players, 5 games, average win ratio: 1

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMRDDEHH/RCRRRCRR
3 players, 6 games, average win ratio: 1

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMDCCEHH/RRRRRDRR
1 player (robinson), 9 games, average win ratio: 1

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////CHRDEDMH/RRRRRCRR
3 players, 9 games, average win ratio: 0.944444

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////CHDCEDMH/RRRRRRRR
4 players, 7 games, average win ratio: 0.916667

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHCDECMH/RRRRDRRR
5 players, 20 games, average win ratio: 0.8875

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHCMERHR/RDRRDRCR
3 players, 34 games, average win ratio: 0.791667

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RMDDECHH/RRRRCRRR
6 players, 11 games, average win ratio: 0.75

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////DHCCEDMH/RRRRRRRR
3 players, 66 games, average win ratio: 0.690476

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHDHECMR/RRRCDRRR
6 players, 17 games, average win ratio: 0.689394

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHCMEDHR/RRRCDRRR
10 players, 17 games, average win ratio: 0.65

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHCDEDMH/RRRRRCRR
7 players, 107 games, average win ratio: 0.62782

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////DHCDECMH/RRRRRRRR
9 players, 34 games, average win ratio: 0.625731

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////DHRMERHD/RCRRRRCR
12 players, 28 games, average win ratio: 0.597222

http://arimaa.com/arimaa/notconv/old/boardimg.php?orient=n&ranks=//////RHDMECHR/RRRCDRRR
56 players, 276 games, average win ratio: 0.577366
Title: Re: Thoughts on my standard set up
Post by 99of9 on Jan 8th, 2015, 6:26am
Interesting analysis.  The adaptive opening book in the old gnobot scored a setup according to:

Sum(result - result_expectation)

where the result_expectation was according to gameroom ratings.  I used the sum rather than the average to ensure it didn't get fooled by one off results.


Arimaa Forum » Powered by YaBB 1 Gold - SP 1.3.1!
YaBB © 2000-2003. All Rights Reserved.