The Winner of NNUE CS IX: Stockfish - Congratulations to SF team!
1) Engines are played via Balsa 031221, Perfect 2022 book: Exception: Houdini 6.03 is played via Cevdet2 bin book
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3764.8 380.0 480 79.2% 2 Komodo Dragon 2.5 : 3711.7 352.0 480 73.3% 3 Fire 8NN.MC3 : 3589.0 276.0 480 57.5% 4 SlowChess 2.8 : 3553.4 252.0 480 52.5% 5 Houdini 6.03 CE : 3540.2 243.0 480 50.6% 6 Koivisto 7.5 : 3510.8 223.0 480 46.5% 7 Berserk 7 NN : 3503.5 218.0 480 45.4% 8 RubiChess 2.2 : 3490.2 209.0 480 43.5% 9 Nemorino 6.0 : 3425.0 166.0 480 34.6% 10 Igel 3.0.5 : 3421.9 164.0 480 34.2% 11 Arasan 23.2 : 3410.8 157.0 480 32.7%
---------------------------------------------------------------
2) Engines are played without using any opening book: Exception: Houdini 6.03 is played via Cevdet2 bin book
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3805.1 321.0 400 80.3% 2 Komodo Dragon 2.5 : 3749.0 297.5 400 74.4% 3 Fire 8NN.MC3 : 3623.5 235.0 400 58.8% 4 SlowChess 2.8 : 3602.4 223.0 400 55.8% 5 Houdini 6.03 CE : 3578.6 210.5 400 52.6% 6 Berserk 7 NN : 3527.0 182.0 400 45.5% 7 Koivisto 7.5 : 3519.9 178.0 400 44.5% 8 RubiChess 2.2 : 3516.3 176.0 400 44.0% 9 Igel 3.0.5 : 3429.0 129.5 400 32.4% 10 Nemorino 6.0 : 3425.0 127.5 400 31.9% 11 Arasan 23.2 : 3408.8 120.0 400 30.0%
---------------------------------------------------------------
3) Both Rankings are merged in one (Books + without Books):
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3782.5 701.0 880 79.7% 2 Komodo Dragon 2.5 : 3728.1 649.5 880 73.8% 3 Fire 8NN.MC3 : 3604.3 511.0 880 58.1% 4 SlowChess 2.8 : 3575.2 475.0 880 54.0% 5 Houdini 6.03 CE : 3557.3 453.5 880 51.5% 6 Koivisto 7.5 : 3514.8 401.0 880 45.6% 7 Berserk 7 NN : 3513.9 400.0 880 45.5% 8 RubiChess 2.2 : 3501.8 385.0 880 43.8% 9 Igel 3.0.5 : 3425.0 293.5 880 33.4% 10 Nemorino 6.0 : 3425.0 293.5 880 33.4% 11 Arasan 23.2 : 3409.9 277.0 880 31.5%
---------------------------------------------------------------
Some Notes: Main target is that, to see the influence with / without opening books Without books: many engines prefer to play almost same openings... But via Balsa, Perfect 2022: much more various openings played Plus it's clear that some NNUE engines suffer without using books Where via Balsa, Perfect book: mostly of engines perform better! To be sure, check please Stockfish's performance as Without Books To be more clear, Stockfish would not gain these higher points... Not so clear? then Houdini 6.03 performance can be compared too E.g Houdini vs no books gained more points...it is another proof that Balsa, Perfect books are stronger than the engines with no books And one thing more (for anyone missed): Without Books: many engines produce lower draw percentage, where Via Balsa suite + Perfect 2022 book, the draws are higher, approx. 5% But in reality... what does it mean? I think that the answer is quite simple: Balsa+Perfect book seems to be stronger than many of the openings Than the engines, which played as without using any opening books... And that's why the draw percentage is higher via Balsa+Perfect books! Actually this is known issue since a log time, but I realized to mention... Btw,.I can work and release such openings, where mostly games can Be ended as win...but then, the error bar margin can be too high..sorry Plus my goal over these years, Perfect books to be better, stronger...! But definitely to not make weaker, the Experts have to agree with Me!) In other words, you may know, The critical, weak openings produce lower draw percentage for sure ! Same words are going for weak engines too, they produce lower draws... Simply the draw issue is just related to engine / opening strength...! Just saying...no more no less...and hope helps!
Continuing....
Meanwhile,a few words about determining any chess engine strength: In case of testing the engines via critical,weak openings, then thousands of Games are required (per player). but using many various openings as well Where via Balsa, Perfect books, minus plus 500-1000 games are enough! Note also that my created openings are not so deep, mainly up to 8 moves What I can say more about Balsa 031221, Perfect 2022 books: 1st of all, Both players (White/Black) will be more equal in opening strength That means, as far as possible, I try to exclude the disadvantage openings 2nd You will save time, no need to run thousands of handicapped games 3rd No doubt, you'll will cut your electricity bills and save money at same time 4th The played games are expecting to be stronger, better in quality for chess 5th You will have chances to see the best/strongest trend openings of all time 6th I work over the openings since 2000 (more than 20 years) and all for free!
And without to not mention this I can't , I wish to say, thanks to all Balsa and Perfect opening book users... Without your interest... my created opening books would not exist! Note that Balsa and Perfect 2022 book's release date is not clear I still work over the strong openings, but probably after a few months... And please stay tuned, and I think it will be worth to wait...because, You'll have one of the most powerful books (mainly up to 8 moves) Sure, my goal is not to beat the strong deep books, it's impossible! Btw, just imagine 48 kg vs 120 kg wrestler...what can be the result ?) But what about in case of test let's say via various short openings?) Then, I have difficulties to find stronger than my books..e.g XXXVII If you are interested in more...I can publish, just let me know please And in case of determining any engine strength, it's not bad to try! That's all for right now, for any further information, please Contact
More Details: All of the Top engines played as NNUE (with their default EvalFiles) Exc. Houdini 6.03 is NON-NNUE and used to play via Cevdet2-BIN In short, for more fair conditions: Houdini played with Deep Book Otherwise, would be much unfair, and now Houdini gains 100+ Elo... Btw, Houdini 6.03's Elo (via short openings): close to Nemorino 6 For more info about why Houdini is played via Cevdet book: Details The Rankings calculated by Ordo, Nemorino 6 is fixed to 3425 Elo Download all the played games (in PGN format): NNUE CS IX
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3764.8 380.0 480 79.2% 2 Komodo Dragon 2.5 : 3711.7 352.0 480 73.3% 3 Fire 8NN.MC3 : 3589.0 276.0 480 57.5% 4 SlowChess 2.8 : 3553.4 252.0 480 52.5% 5 Houdini 6.03 CE : 3540.2 243.0 480 50.6% 6 Koivisto 7.5 : 3510.8 223.0 480 46.5% 7 Berserk 7 NN : 3503.5 218.0 480 45.4% 8 RubiChess 2.2 : 3490.2 209.0 480 43.5% 9 Nemorino 6.0 : 3425.0 166.0 480 34.6% 10 Igel 3.0.5 : 3421.9 164.0 480 34.2% 11 Arasan 23.2 : 3410.8 157.0 480 32.7%
- Balsa+Perfect book's overall draw perc. (based on 2640 games): 58%
---------------------------------------------------------------
2) Engines are played without using any opening book: Exception: Houdini 6.03 is played via Cevdet2 bin book
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3805.1 321.0 400 80.3% 2 Komodo Dragon 2.5 : 3749.0 297.5 400 74.4% 3 Fire 8NN.MC3 : 3623.5 235.0 400 58.8% 4 SlowChess 2.8 : 3602.4 223.0 400 55.8% 5 Houdini 6.03 CE : 3578.6 210.5 400 52.6% 6 Berserk 7 NN : 3527.0 182.0 400 45.5% 7 Koivisto 7.5 : 3519.9 178.0 400 44.5% 8 RubiChess 2.2 : 3516.3 176.0 400 44.0% 9 Igel 3.0.5 : 3429.0 129.5 400 32.4% 10 Nemorino 6.0 : 3425.0 127.5 400 31.9% 11 Arasan 23.2 : 3408.8 120.0 400 30.0%
- No Book's overall draw percentage (based on 2200 games): 53%
---------------------------------------------------------------
3) Both Rankings are merged in one (Books + without Books):
# PLAYER : RATING POINTS GAMES (%) 1 Stockfish 011221 : 3782.5 701.0 880 79.7% 2 Komodo Dragon 2.5 : 3728.1 649.5 880 73.8% 3 Fire 8NN.MC3 : 3604.3 511.0 880 58.1% 4 SlowChess 2.8 : 3575.2 475.0 880 54.0% 5 Houdini 6.03 CE : 3557.3 453.5 880 51.5% 6 Koivisto 7.5 : 3514.8 401.0 880 45.6% 7 Berserk 7 NN : 3513.9 400.0 880 45.5% 8 RubiChess 2.2 : 3501.8 385.0 880 43.8% 9 Igel 3.0.5 : 3425.0 293.5 880 33.4% 10 Nemorino 6.0 : 3425.0 293.5 880 33.4% 11 Arasan 23.2 : 3409.9 277.0 880 31.5%
---------------------------------------------------------------
Some Notes: Main target is that, to see the influence with / without opening books Without books: many engines prefer to play almost same openings... But via Balsa, Perfect 2022: much more various openings played Plus it's clear that some NNUE engines suffer without using books Where via Balsa, Perfect book: mostly of engines perform better! To be sure, check please Stockfish's performance as Without Books To be more clear, Stockfish would not gain these higher points... Not so clear? then Houdini 6.03 performance can be compared too E.g Houdini vs no books gained more points...it is another proof that Balsa, Perfect books are stronger than the engines with no books And one thing more (for anyone missed): Without Books: many engines produce lower draw percentage, where Via Balsa suite + Perfect 2022 book, the draws are higher, approx. 5% But in reality... what does it mean? I think that the answer is quite simple: Balsa+Perfect book seems to be stronger than many of the openings Than the engines, which played as without using any opening books... And that's why the draw percentage is higher via Balsa+Perfect books! Actually this is known issue since a log time, but I realized to mention... Btw,.I can work and release such openings, where mostly games can Be ended as win...but then, the error bar margin can be too high..sorry Plus my goal over these years, Perfect books to be better, stronger...! But definitely to not make weaker, the Experts have to agree with Me!) In other words, you may know, The critical, weak openings produce lower draw percentage for sure ! Same words are going for weak engines too, they produce lower draws... Simply the draw issue is just related to engine / opening strength...! Just saying...no more no less...and hope helps!
Continuing....
Meanwhile,a few words about determining any chess engine strength: In case of testing the engines via critical,weak openings, then thousands of Games are required (per player). but using many various openings as well Where via Balsa, Perfect books, minus plus 500-1000 games are enough! Note also that my created openings are not so deep, mainly up to 8 moves What I can say more about Balsa 031221, Perfect 2022 books: 1st of all, Both players (White/Black) will be more equal in opening strength That means, as far as possible, I try to exclude the disadvantage openings 2nd You will save time, no need to run thousands of handicapped games 3rd No doubt, you'll will cut your electricity bills and save money at same time 4th The played games are expecting to be stronger, better in quality for chess 5th You will have chances to see the best/strongest trend openings of all time 6th I work over the openings since 2000 (more than 20 years) and all for free!
And without to not mention this I can't , I wish to say, thanks to all Balsa and Perfect opening book users... Without your interest... my created opening books would not exist! Note that Balsa and Perfect 2022 book's release date is not clear I still work over the strong openings, but probably after a few months... And please stay tuned, and I think it will be worth to wait...because, You'll have one of the most powerful books (mainly up to 8 moves) Sure, my goal is not to beat the strong deep books, it's impossible! Btw, just imagine 48 kg vs 120 kg wrestler...what can be the result ?) But what about in case of test let's say via various short openings?) Then, I have difficulties to find stronger than my books..e.g XXXVII If you are interested in more...I can publish, just let me know please And in case of determining any engine strength, it's not bad to try! That's all for right now, for any further information, please Contact
More Details: All of the Top engines played as NNUE (with their default EvalFiles) Exc. Houdini 6.03 is NON-NNUE and used to play via Cevdet2-BIN In short, for more fair conditions: Houdini played with Deep Book Otherwise, would be much unfair, and now Houdini gains 100+ Elo... Btw, Houdini 6.03's Elo (via short openings): close to Nemorino 6 For more info about why Houdini is played via Cevdet book: Details The Rankings calculated by Ordo, Nemorino 6 is fixed to 3425 Elo Download all the played games (in PGN format): NNUE CS IX