How close to Human FIDE rating are the engine ratings?

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

Moderators: hgm, Rebel, chrisw

User avatar
Ovyron
Posts: 4556
Joined: Tue Jul 03, 2007 4:30 am

Re: How close to Human FIDE rating are the engine ratings?

Post by Ovyron »

lkaufman wrote: Sun Jun 21, 2020 3:54 am Do you have a link for the best estimates of the Stockfish skill levels relative to CCRL 40/15 rating list?
This is the thread you might be looking for. Ferdy did extensive testing of UCI_ELO and UCI_LimitStrength features of many engines, not just Stockfish.

It seems you have commented on that thread, so yeah, that's as far as testing has gotten.

Note that it seems MikeB's series of engines with Honey's code can be more sophisticated and accurate than Stockfish's default implementation.

Here's how this was done:

MikeB wrote: Tue Dec 31, 2019 7:02 pm With Honey , rating of 1712 or so was anchored to Shallow Blue (CCRL rating of 1712 or so ). After it was all done , I then realized CCRL at lower ratings is much stronger than FIDE - CCRL 1700 is 2000+ FIDE , so Kai had posted a formula to convert from FIDE to CCRL , so I reversed the formula to go from CCCR to FIDE ( so users can Play FIDE Elo and not CCRL Elo). CCRL 2800 is about 2800 FIDE above 2800 , FIDE is stronger , I.e., 2900 FIDE is stronger than 2900 CCRL.
If this formula is correct, what this thread has been talking about has already been achieved since then. We'd just need to tweak the formula.