Abalone BGA ELO vs Abalone App and the Play Abalone website

[CameronHamill]

Abalone BGA ELO vs Abalone App and the Play Abalone website

Hi Members,

I'm just wondering what you think a reasonable comparison is between Abalone BGA ELO vs Abalone App and the Play Abalone website?

I have asked a similar question in relation to the estimated ELO of the Patchwork game APP Medium AI Bot vs Human Players on BGA.

Essentially what I'm interested in opinions on is this;

The Belgian Daisy Opening Configuration on the Abalone App against Hard AI vs the ELO of a given Human player on BGA.

Also the Hard AI Bot and the Expert AI Bot with the Belgian Daisy Opening Configuration on the Play Abalone website vs a given Human player on BGA.

Why am I asking the question? I'm just trying to gauge my approximate skill level. For example I can beat Hard AI on the Abalone app in the Belgian Daisy Opening Configuration 6 points to 5 points occasionally. But I have a feeling that it's no match for beginner levels of Abalone Belgian Daisy on BGA. Essentially I'm trying to gauge at which point I'll be able to compete effectively against human players. Would I need to beat Hard AI 6 to 0 to stand a chance on BGA?

I have played around 136 matches mostly against Hard AI on the app with the Belgian Daisy opening and have won around 40 matches. Usually only 6 to 5. I feel that perhaps if I can start winning 6 to 3 or 6 to 2 I may be competitive on BGA against average players.

I wonder what you would consider are approximations for actual BGA ELO levels and these Hard and Expert AI bots?

[Ceaseless]

Haven't checked for Abalone specifically, but this is a common song and dance from Chess. An elo pool compares players within the pool, it's not designed to cross compare numbers from different sites/pools. Especially with how inactive Abalone's pool seems to be on BGA. Trying to cross compare computers and human play is even worse. Computers don't play like people, they make good moves, then randomly insert bad ones for their level. They don't have brains to regulate play and act like any sort of rating you'd see from human play, where genuine mistakes are made with actual reasons. I'd be surprised to see any deviation from how it works from Chess in Abalone, there are probably more issues with Abalone due to the activity gap.

If you want to know how you'd fare vs actual players, play them and find out.

[six billes]

The ELO ranking on BGA is completely distorted.
Most truly good players have lower ELOs than those at the top of the ELO ranking scale.
This is because the ELO ranking shouldn't take into account player games outside of competitive mode.
For example: Six Billets or Chadmild play mainly with beginners, but they play a lot of games, and over time, they accumulate a lot of points. However, they are less good at this game than Julienaba or Talh, who play far fewer games and almost exclusively with other good players like themselves.
My conclusion is that the ELO ranking should be removed from the "single game" mode and kept only for the "Arena" and "Tournament" modes.
If we want this to mean anything on BGA.

[CameronHamill]

Hi Ceaseless,

Thanks very much for replying. I see exactly what you're saying that it's basically an Apple vs Oranges comparison. In the sense that you can only gauge your performance against humans by playing humans. And that playing against AI will only show your performance against AI.

I suppose my style is I like to try and practice AI games before 'going live' against human players for example. I find with a game like Abalone or Patchwork I'm trying to develop a sort of baseline for myself. Pretty much to not make a fool of myself against human players so to speak and to not be easily beaten.

I take all your points which have definitely changed my thoughts on human vs AI opponents.

Hi six billes,

Thanks very much for your reply. I see exactly what you're saying in relation to BGA ELO. About how there are different approaches in how Abalone players play the game. You revealed to me some new insights about how to look at ELO.

As a side note would you tend to play with the Belgian Daisy Opening I wonder? Do you have a favourite opening configuration?

I very much appreciate this game.

I can beat Hard AI in Belgian Daisy Opening Configuration occasionally with 6 to 5 and sometimes 6 to 4 on the Abalone Android App.

On Play Abalone (dot) com I tested myself today and lost 2 or 3 times to Medium AI in the Belgian Daisy Opening Configuration. I lost around 6 to 4 in most of the matches. So it seems I have a lot more practice to do. And that Hard AI on the Abalone Android App isn't hard at all. It's possibly only actually easy on Play Abalone (dot) com in my guess.

[six billes]

Do you have a favourite opening configuration?

swiss daisy

https://abaloneonline.wordpress.com/wp- ... -daisy.png

[CameronHamill]

That's a good call six billes!

We must try to represent our countries

Now we need an Australian Daisy

[thewarlordess]

Hi the bot on playabalone is dreadful. It literally moves single marbles in front of yours so you can push them over edge. Worst bot I ever played.
By the way can we please have more openings? There are so many cool ones....https://abaloneonline.wordpress.com/wp- ... 0x1025.jpg