Engine's "ELO" revisited

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Rebel, chrisw

bob
Posts: 20943
Joined: Mon Feb 27, 2006 7:30 pm
Location: Birmingham, AL

Re: Engine's "ELO" revisited

Post by bob »

Ponti wrote: Sun Dec 15, 2019 5:00 pm Most of the time I play 5 min+ 3 sec./move time increment.
Last 8 games I lost 7-1 to Prodeo 2.4. :shock:

I also want to try Bob's Crafty 25.3 , but I think it is hard to set the crafty.rc file to a similar level.

Shredder's adaptive function is good, I'm playing hard games with an old Android tablet (Motorola Xoom, 1st edition).

Things like tablebases and opening books also matters here. It's not fair to reach an endgame against an engine with tablebases, so I almost all the time turn them off.

I don't know why, but some engines without opening book choose 1...Nc6 against 1.e4, so maybe that's the best reply to it... :lol:
Try the most recent version, 25.4... It has an "elo" command. If you say "Elo 2800" I can't really confirm that it will play at 2800. But it has a basic mapping of nodes per second to elo so that "elo 2600" actually plays about 200 Elo weaker than "elo 2800". Now if I can somehow normalize "elo 2800" to an actual Elo, this could be tuned and ought to work pretty well. Note that it was tuned at fast games, so how it performs at very slow games is not tested. I presume that the overall rating will drop some at longer time controls, although dropping the "elo xxx" to "elo xxx-200" ought to still be pretty close to a 200 point drop.

Tested pretty thoroughly, except for the actual mapping of my "elo setting" to the human scale. Probably should once again set Crafty up on a chess server and let it run, but with a formula that only allows longer time controls. But even that is problematic due to humans using computers to cheat online.
User avatar
Ponti
Posts: 493
Joined: Wed Mar 15, 2006 6:13 am
Location: Curitiba - PR - BRAZIL

Re: Engine's "ELO" revisited

Post by Ponti »

" Now if I can somehow normalize "elo 2800" to an actual Elo, this could be tuned and ought to work pretty well. "

Well, that's easy, you have to invite only +- the top 10 players , because only them are playing at >2800< ELO level. :lol:

1 Carlsen, Magnus

2 Caruana, Fabiano

3 Ding, Liren

4 Vachier-Lagrave, Maxime

5 Grischuk, Alexander

6 Aronian, Levon

7 Mamedyarov, Shakhriyar

8 Giri, Anish

9 Nepomniachtchi, Ian

10 Radjabov, Teimour

You'll have to weaken the engine , because if you don't it will win the tournament (and Magnus will not be happy with that).
A. Ponti
AMD Ryzen 1800x, Windows 10.
FIDE current ratings: standard 1913, rapid 1931
Canoike
Posts: 125
Joined: Tue Jan 17, 2012 8:08 pm

Re: Engine's "ELO" revisited

Post by Canoike »

Shredder from Stefan Meyer-Kahlen has an option to adapt its strength. If you lose, it lowers its rating and the opposite of course. The Gui hosts this feature. This is what I use when I want to play for fun.
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: Engine's "ELO" revisited

Post by carldaman »

bob wrote: Wed Dec 18, 2019 5:20 am
Try the most recent version, 25.4... It has an "elo" command.
Hello Bob, where can v25.4 be obtained?
Thanks,
Carl
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Engine's "ELO" revisited

Post by MikeB »

carldaman wrote: Thu Dec 19, 2019 12:18 am
bob wrote: Wed Dec 18, 2019 5:20 am
Try the most recent version, 25.4... It has an "elo" command.
Hello Bob, where can v25.4 be obtained?
Thanks,
Carl
That would be me - I got distracted for a few days. Will put it on Github in a few minutes.
Image
User avatar
MikeB
Posts: 4889
Joined: Thu Mar 09, 2006 6:34 am
Location: Pen Argyl, Pennsylvania

Re: Engine's "ELO" revisited

Post by MikeB »

MikeB wrote: Thu Dec 19, 2019 5:42 am
carldaman wrote: Thu Dec 19, 2019 12:18 am
bob wrote: Wed Dec 18, 2019 5:20 am
Try the most recent version, 25.4... It has an "elo" command.
Hello Bob, where can v25.4 be obtained?
Thanks,
Carl
That would be me - I got distracted for a few days. Will put it on Github in a few minutes.
Here you go:

https://github.com/MichaelB7/Crafty
Image
User avatar
xr_a_y
Posts: 1871
Joined: Sat Nov 25, 2017 2:28 pm
Location: France

Re: Engine's "ELO" revisited

Post by xr_a_y »

If you give a try to Minic level function from 1 to 100, I d be glad to have feedback.
carldaman
Posts: 2283
Joined: Sat Jun 02, 2012 2:13 am

Re: Engine's "ELO" revisited

Post by carldaman »

MikeB wrote: Thu Dec 19, 2019 6:56 am
MikeB wrote: Thu Dec 19, 2019 5:42 am
carldaman wrote: Thu Dec 19, 2019 12:18 am
bob wrote: Wed Dec 18, 2019 5:20 am
Try the most recent version, 25.4... It has an "elo" command.
Hello Bob, where can v25.4 be obtained?
Thanks,
Carl
That would be me - I got distracted for a few days. Will put it on Github in a few minutes.
Here you go:

https://github.com/MichaelB7/Crafty
Thanks Mike & Bob, enjoy your holidays! :)
ydebilloez
Posts: 163
Joined: Tue Jun 27, 2017 11:01 pm
Location: Lubumbashi
Full name: Yves De Billoëz

Re: Engine's "ELO" revisited

Post by ydebilloez »

Try any of these XBoard/UCI engines:
These should remain bad even at fastest CPU. Start from bottom for weakest.
I got trouble in getting the binaries for a lot of them, good luck.

1: Fimbulwinter 5.02
2: DarkFusch 0.09
3: Qutechess 1.01t rev3 ja
4: Pierre 1.7 ja
5: Kace 0.82 ja
6: Raffaela 0.14
7: Gringo 1.49b
8: Kcchess pre-alpha2 rev 1.13
9: Xadreco 5.7
10: Strategicdeep 1.25
11: Dreamer 0.2.0
12: Delfi 1000 ELO
13: Giuchess 1.0b2dc
14: Butterfly 0.1.7_alpha
15: Nanook 0.17
16: Bigbook 3.1
17: Dika 0.4209
18: Ochess rev2 ja
19: Mfchess 1.3
20: Luzhin 0.1
21: Youk 1.05
22: Belofte 0.2.8
23: Pyotr Am 0.6
24: Cheops 1.1
25: Dendron 0.42b ja
26: Geko 0.43
27: Neg 0.3d
28: Sachy 0.2.01
29: Prechess 0.7.8
30: Microchess1976
31: Tri-OS
32: Ram 2.0
33: Omar 3.1
34: Ace v0.1
35: Neurone I
36: Cpp1
37: Brutus RND
38: Deep Duke 1.00ja
39: NSVChess 0.14
40: Pos-1.20
Yves De Billoëz @ macchess belofte chess
Once owner of a Mephisto I, II, challenger, ... chess computer.
Canoike
Posts: 125
Joined: Tue Jan 17, 2012 8:08 pm

Re: Engine's "ELO" revisited

Post by Canoike »

xr_a_y wrote: Thu Dec 19, 2019 8:53 am If you give a try to Minic level function from 1 to 100, I d be glad to have feedback.
I compiled minic with the syzygy option and got a Linux executable called minic_WITH_SYZYGY_linux_x64. The last line of the JSON file is as follows :
"syzygyPath": "/home/user/jeux/echecs/syzygy/"
1) I must use the UCI protocol in order to be able to start a classic game, Winboard 1 or 2 protocol do not allow to play.
2) I tested the endgame with bishop and knight vs a lonely king. The score is about +99.8 but it cannot mate.