UCI chess engines with low ELO (<1700)

Discussion of chess software programming and technical issues.

Moderator: Ras

rdhoffmann
Posts: 45
Joined: Fri Apr 21, 2023 3:46 pm
Full name: Richard Hoffmann

Re: UCI chess engines with low ELO (<1700)

Post by rdhoffmann »

Ok have to try Crafty, Stockfish skill level makes no sense to me. Seems SF randomly plays with ELO 3000+ even at low skill levels which is not realistic by any means (i.e. if you're unlucky enough, it plays with monster strength). I noticed this setting up tournaments against Pedantic and my own engine, both should be 2800+ and be able to win consistently vs the lower SF skill levels.
David Carteau
Posts: 139
Joined: Sat May 24, 2014 9:09 am
Location: France
Full name: David Carteau

Re: UCI chess engines with low ELO (<1700)

Post by David Carteau »

rdhoffmann wrote: Fri Dec 20, 2024 9:46 am I'm trying to write a weak chess engine (for human play, to have fun with). In order to know roughly the ELO strength, it could be useful to play against other engines.

Does anyone know chess engines rated below 1700 or so, which are:

- stable under UCI protocol
- free
- not always making the same move, esp in the opening

I recall the engines that come with Arena are well above 2000 already and thus far too strong for this purpose.
If you want a (really) weak engine as a "first" (weakest) opponent for your project, you can try Nostradamus, which I released three weeks ago. I have played hundreds of games with it under cutechess-cli without any problems.

Image

The engine is derived from an "experiment" where I try to train a language model to play chess :) The engine is actually trained to predict the next move (i.e. depth = 1) from a given position (no search), but it is possible to tweak the (Python) code to introduce more variety if you are interested !

Regards,

David
Download the Orion chess engine --- Train your NNUE with the Cerebrum library --- Contribute to the Nostradamus experiment !
IanKennedy
Posts: 56
Joined: Sun Feb 04, 2018 12:38 pm
Location: UK

Re: UCI chess engines with low ELO (<1700)

Post by IanKennedy »

TSCP is reliable and in that ELO range although it uses winboard. It runs in all the GUIs I use.

http://www.tckerrigan.com/Chess/TSCP/
Author of the actively developed PSYCHO chess engine
User avatar
Ras
Posts: 2671
Joined: Tue Aug 30, 2016 8:19 pm
Full name: Rasmus Althoff

Re: UCI chess engines with low ELO (<1700)

Post by Ras »

rdhoffmann wrote: Fri Dec 20, 2024 9:46 am- not always making the same move, esp in the opening
The opening book is usually meant to be controlled by the GUI, so just configure your GUI to get a variety of openings.
Rasmus Althoff
https://www.ct800.net
User avatar
Ajedrecista
Posts: 2042
Joined: Wed Jul 13, 2011 9:04 pm
Location: Madrid, Spain.

Re: UCI chess engines with low ELO (< 1700).

Post by Ajedrecista »

Hello Richard:

I read some years ago that Faile is a very stable engine. Faile 1.4.4 is rated currently at 1930 Elo at CCRL Blitz and 1957 Elo at CCRL 40/15. The download page of Faile is:

https://faile.sourceforge.net/download.php

With links still working given that the engine was released in 2000! OTOH, the engine is WinBoard and I do not know if it works with modern Windows versions. Good luck with your project!

Regards from Spain.

Ajedrecista.
User avatar
Tibono
Posts: 122
Joined: Sat Aug 01, 2015 6:16 pm
Location: France
Full name: Eric Bonneau

Re: UCI chess engines with low ELO (<1700)

Post by Tibono »

Hi,
Kenny_Lichess_1600, also _1200, are worth considering. These are specific eval files to be paired with Stockfish 12.
I use nodestime 200 for Kenny_Lichess_1600 and 30 for Kenny_Lichess _1200.
Tibono
User avatar
hgm
Posts: 28268
Joined: Fri Mar 10, 2006 10:06 am
Location: Amsterdam
Full name: H G Muller

Re: UCI chess engines with low ELO (<1700)

Post by hgm »

Most engines in that Elo range use WB protocol. Some of those are very stable. Of course they have all kinds of bugs, but not all bugs result in crashes or time forfeits.

http://www.open-aurec.com/chesswar/Ches ... 017PLs.htm
rdhoffmann
Posts: 45
Joined: Fri Apr 21, 2023 3:46 pm
Full name: Richard Hoffmann

Re: UCI chess engines with low ELO (<1700)

Post by rdhoffmann »

Thank you guys for all the great answers! :D
TSCP worked out of the box with zero issues, and with the opening book idea from Ras it is a perfect sparring partner.

I noticed that with just q-search (no move ordering, no pruning, no hash table) I'm getting close to the desired ELO target already.

Will check the other recommendations as well.