FIDE World Championship 2021 analyzed by 10 Top engines!

[AlexChess]

Magnus Carlsen (2855) - Ian Nepomniachtchi (2782)

The format involves a two-player match over 14 standard games. The thinking time is 120 minutes for the first 40 moves, followed by 60 minutes for the next 20 moves, and then 15 minutes for the remainder of the game with an increase of 30 seconds per move starting at move 61. In case of parity 7-7, a rapid / blitz tiebreak will be played.

We will deeply analyze the best positions from Nov, 26 to Dec. 16 2021

Fat Titz 1.1 by Sopel & Stockfish Team 310921 + M14.3 book (Cfish derivative + huge NNUE by Sopel) 3731 ELO
SwordFish 14.4 (by Bookmaker & Stockfish Team) (optimized for tactic) 3683 ELO
Shashchess 20 (by Andrea Manzo, Omar Khalid & Stockfish Team) 3679 ELO
BrainLearn 15 (by Andrea Manzo, Omar Khalid & Stockfish Team) 3679 ELO
Bluefish NN 220921 (by Bookmaker & Stockfish Team) 3678 ELO
SF-PB 291021 (by Chessman & Stockfish Team, compiled by Chessman + M14.3 book - Stockfish dev + polyglot book) 3677 ELO
Blue Marlin 14.4a (by Bookmaker & Stockfish Team, optimized for sacrifices) 3664 ELO
Fat Fritz 2.6 in Stockfish 14.1 231921 github (by Stockfish Team & NNUE by Albert silver) 3656 ELO
Sugar AI ICCF (by Marco Zerbinati, Omar Khalid & Stockfish Team, optimized for long time analysis) 3654 ELO
CorChess 2.0 NN 150121 (by Bookmaker & Stockfish Team, optimized for long time analysis) 3644 ELO

Watch all the games LIVE here: https://www.chess.com/events/2021-fide- ... ampionship

[Jouni]

Better to say: 1 Top engine = Stockfish .

[AlexChess]

Better to say: 1 Top engine = Stockfish .

Stockfish and derivatives are still 100-150 ELO over the best original engines like Dragon 2.5.1, Rubichess 2.2, Berserk 7 and Koivisto 7.5. But luckily for diversity, not SF derivatives are becoming every day stronger!

Note: CorChess 2 NN is compiled by Ivan Ilvec, not by Bookmaker, Fat Fritz 2.6 in Stockfish 14.1 231021

[Vinvin]

Better to say: 1 Top engine = Stockfish .

Stockfish and derivatives are still 100-150 ELO over the best original engines like Dragon 2.5.1, Rubichess 2.2, Berserk 7 and Koivisto 7.5. But luckily for diversity, not SF derivatives are becoming every day stronger!

Note: CorChess 2 NN is compiled by Ivan Ilvec, not by Bookmaker, Fat Fritz 2.6 in Stockfish 14.1 231021

Sometimes, some derivatives are way stronger than Stockfish because Stockfish is completely blind to some tactics/understanding.
But personally, I would only use one derivative : Blue Marlin 14.4a because it's the engine who solves the most of special positions.

[dkappe]

Better to say: 1 Top engine = Stockfish .

Stockfish and derivatives are still 100-150 ELO over the best original engines like Dragon 2.5.1…

100 you say? Hmmm.

SF14.1/Dragon 2.5 Elo delta:
CEGT 40/4 1CPU: 56
Stefan Pohl: 49
Fast GM 10+6: 41
Rebel Gambit List: 62

But what’s 50 elo between friends?

[Uri Blass]

Better to say: 1 Top engine = Stockfish .

Stockfish and derivatives are still 100-150 ELO over the best original engines like Dragon 2.5.1, Rubichess 2.2, Berserk 7 and Koivisto 7.5. But luckily for diversity, not SF derivatives are becoming every day stronger!

Note: CorChess 2 NN is compiled by Ivan Ilvec, not by Bookmaker, Fat Fritz 2.6 in Stockfish 14.1 231021

Stockfish is not 100-150 elo stronger than Dragon2.5.1

https://ccrl.chessdom.com/ccrl/4040/

Stockfish 14 64-bit 4CPU 3547
Dragon by Komodo 2.5 64-bit 4CPU 3518

[AlexChess]

True for official SF 14.1 , but Fat Tits 1.1 (derivative of a more recent SF-dev source) is +100 ELO. Test It, too

https://github.com/Sopel97/FatTitz/releases/tag/v1.1

[dkappe]

True for official SF 14.1 , but Fat Tits 1.1 (derivative of a more recent SF-dev source) is +100 ELO. Test It, too

https://github.com/Sopel97/FatTitz/releases/tag/v1.1

Then it must be 50 elo over SF14.1. Released August 2021, while SF 14.1 was released in late October. I’m surprised they didn’t release it instead of 14.1. Not on any of the rating lists? Must be a conspiracy.

[mehmet123]

Better to say: 1 Top engine = Stockfish .

Stockfish and derivatives are still 100-150 ELO over the best original engines like Dragon 2.5.1…

100 you say? Hmmm.

SF14.1/Dragon 2.5 Elo delta:
CEGT 40/4 1CPU: 56
Stefan Pohl: 49
Fast GM 10+6: 41
Rebel Gambit List: 62

But what’s 50 elo between friends?

According to one to one matches elo difference is 10-15 elo with some special settings for Dragon.
forum3/viewtopic.php?p=907966#p907966

[AlexChess]

True for official SF 14.1 , but Fat Tits 1.1 (derivative of a more recent SF-dev source) is +100 ELO. Test It, too

https://github.com/Sopel97/FatTitz/releases/tag/v1.1

Then it must be 50 elo over SF14.1. Released August 2021, while SF 14.1 was released in late October. I’m surprised they didn’t release it instead of 14.1. Not on any of the rating lists? Must be a conspiracy.

If they does not test also Fat Tits 1.1 (deeply modified Cfish by Sopel + his huge NNUE) SF-PB perfectly compiled by Chessman from latest SF-dev, Swordfish 14.4, Blue Marlin 14.4a, FatFritz in Stockfish 14.1 + Albert NNUE, Night Nurse NN updated with last compatible Cfish) LIKE I DO, they never know how Stockfish 14.1 is drawish and FLAT related to modified derivatives.

...But they continue to test illegal Houdini, Fat Fritz 2 Chessbase...

PS: I hope that later today you will also analyse here first World Championship 2021 game



...And you will release new huge NNUEs like Sopel