Latest Update: 25.02.2021
Hello Chess Friends,
As usual, I am pleased to announce that,Recently I organized a new book competition!This kind of tournament is different than all...
Some Details,All deep books are played by Houdini 6.03 Pro x64-pextWhich is NON-NNUE and Houdini 6.03 is close to 3500 EloPlus, Cevdet2-BIN book is tested via Komodo 13.2.5 tooWhere the opponents are Engines+EvalFile+Perfect 2021In short, my main target was that,NON-NNUE (av. 3500 Elo) vs NNUE (av.3300-3700 Elo)E.g NON-NNUE+Deep openings vs NNUE+Short openings...
To be more clear,The idea of running this kind of tournament is born due to,NNUE / Evalfiles are with clear advantage over NON-NNUE...For example, as I said before in my recent organized DuelMany testers prefer short books for NON-NNUE vs NNUE....E.g NNUE engines gain in deep openings, middle game etc.Where NON-NNUE gains mainly up to 5-8 moves....So...running with similar conditions: is it much fair...?...To be honest.., I do not think so....At least with the new tourney...we will have chances to seeNON-NNUE engines under better and more fair conditions...Because, NON-NNUE engines will be played via deep booksExc. just for comparing, Houdini6.03 via Perfect2021 played tooIn other words,Now it's just opposite....NON-NNUE via deep opening lines...Where NNUE engines are used to play via Perfect 2021 book
Another note is that, about to not affecting book learning of CS L,New/separate fresh book copies are installed...(for current tour..)Meanwhile, just I'd like to add,Perfect 2021 book is not optimized over similar book matches ...So no BIG surprise that any well-tuned deep book performs better...Sure, in case of using close in strength engine for all books!)Btw, the reality is reality, actually this is already known...For example, in case of various openings e.g up to 8 moves:Then usually Perfect books''s performance is one of best!As I stated before (for anyone who missed about Perfect2021):Not so deep in opening lines, just mainly up to 8 moves!Recommended to be used as same book for all chess engines...And it can be a good choice to determine any engine strength!Plus Perfect2021 prefers to play a lot of various openings...What does it mean in chess? a long story...but I will try in short:The strongest opening theory is quite limited, not so various!In other words, I can increase the strength of Perfect 2021 bookBut then, I'm afraid we will see minus plus 10-15 main openings!Plus in that case, the engine testings would not be much boring?!
A few more words about the current played deep opening books,They are not so much various in openings...their target is very simple:To win...and all of them prefer mainly the strongest trend openings!That sounds perfectly fine to Me.. the best way to get more points!Here are final standings, frankly I'm quite impressed by Cevdet books!E.g approx.80-90 Elo better than Perfect 2021 book... just unbelievable!That means also, in overall openings Cevdet books are superior strong!!Plus, Cerebellum, Katzenmaier, DON, TrendMoves etc performed very well too!
CS LI's Champion: Cevdet2-BIN - Congratulations to Cevdet Sarı!
# PLAYER RATING POINTS GAMES (%) 1 Stockfish 170121 : 3712.9 1004.5 1370 73.3% 2 Eman 6.91 : 3712.3 934.0 1250 74.7% 3 CF EXT 161120 : 3708.2 852.5 1170 72.9% 4 Fat Fritz 2 : 3675.4 799.0 1150 69.5% 5 Komodo Dragon : 3650.0 896.5 1370 65.4% 6 Cevdet2-BIN HD : 3563.8 154.5 300 51.5% 7 Perfect2 Cevdet HD : 3558.3 152.5 300 50.8% 8 Cerebellum 3M Dual HD : 3555.5 151.5 300 50.5% 9 Katzenmaier1701 HD : 3554.1 151.0 300 50.3% 10 DON Lite 3 HD : 3543.3 147.0 300 49.0% 11 TrendMoves2020 HD : 3541.5 146.5 300 48.8% 12 1337chess 041020 HD : 3537.3 145.0 300 48.3% 13 Cevdet2-BIN KO : 3536.0 144.5 300 48.2% 14 RadaR 050121 HD : 3536.0 144.5 300 48.2% 15 M11.3 HD : 3536.0 144.5 300 48.2% 16 Alchemist 110121 HD : 3531.8 143.0 300 47.7% 17 HurnaTal v2 HD : 3523.4 140.0 300 46.7% 18 2600-Plus-Club HD : 3523.4 140.0 300 46.7% 19 TR Power HD : 3515.0 137.0 300 45.7% 20 NimaTiv Beta HD : 3509.3 135.0 300 45.0% 21 Perfect 2021 HD : 3472.3 406.5 1000 40.6% 22 SlowChess 2.5 : 3458.8 565.5 1450 39.0% 23 Nemorino 6.00 : 3453.9 556.5 1450 38.4% 24 RubiChess 2.0.1 : 3396.5 452.0 1450 31.2% 25 Minic 2.53 : 3370.3 393.5 1350 29.1% 26 Igel 2.9.0 : 3370.1 393.0 1350 29.1%
• Calculated by Ordo chess program / KO = Komodo 13.2.5 / HD = Houdini 6.03
More Details: • Red Name = Deep books, played by same chess engine: Houdini 6.03 Exception, The Champion (Cevdet2) is tested via Komodo 13.2.5 too • Black Name = Engines are used to play via Perfect 2021 opening book • As far as possible, as book participants are played Top ranked of CS L • All opening books are used to play by Author's recommended settings • New book entries are played min 300 games per player (vs NNUE...) • The overall draw percentage (based on 9430 games) not so high: 53% • Download all of the played tour games (in PGN format): SCCT CS LI
Hello Chess Friends,
As usual, I am pleased to announce that,Recently I organized a new book competition!This kind of tournament is different than all...
Some Details,All deep books are played by Houdini 6.03 Pro x64-pextWhich is NON-NNUE and Houdini 6.03 is close to 3500 EloPlus, Cevdet2-BIN book is tested via Komodo 13.2.5 tooWhere the opponents are Engines+EvalFile+Perfect 2021In short, my main target was that,NON-NNUE (av. 3500 Elo) vs NNUE (av.3300-3700 Elo)E.g NON-NNUE+Deep openings vs NNUE+Short openings...
To be more clear,The idea of running this kind of tournament is born due to,NNUE / Evalfiles are with clear advantage over NON-NNUE...For example, as I said before in my recent organized DuelMany testers prefer short books for NON-NNUE vs NNUE....E.g NNUE engines gain in deep openings, middle game etc.Where NON-NNUE gains mainly up to 5-8 moves....So...running with similar conditions: is it much fair...?...To be honest.., I do not think so....At least with the new tourney...we will have chances to seeNON-NNUE engines under better and more fair conditions...Because, NON-NNUE engines will be played via deep booksExc. just for comparing, Houdini6.03 via Perfect2021 played tooIn other words,Now it's just opposite....NON-NNUE via deep opening lines...Where NNUE engines are used to play via Perfect 2021 book
Another note is that, about to not affecting book learning of CS L,New/separate fresh book copies are installed...(for current tour..)Meanwhile, just I'd like to add,Perfect 2021 book is not optimized over similar book matches ...So no BIG surprise that any well-tuned deep book performs better...Sure, in case of using close in strength engine for all books!)Btw, the reality is reality, actually this is already known...For example, in case of various openings e.g up to 8 moves:Then usually Perfect books''s performance is one of best!As I stated before (for anyone who missed about Perfect2021):Not so deep in opening lines, just mainly up to 8 moves!Recommended to be used as same book for all chess engines...And it can be a good choice to determine any engine strength!Plus Perfect2021 prefers to play a lot of various openings...What does it mean in chess? a long story...but I will try in short:The strongest opening theory is quite limited, not so various!In other words, I can increase the strength of Perfect 2021 bookBut then, I'm afraid we will see minus plus 10-15 main openings!Plus in that case, the engine testings would not be much boring?!
A few more words about the current played deep opening books,They are not so much various in openings...their target is very simple:To win...and all of them prefer mainly the strongest trend openings!That sounds perfectly fine to Me.. the best way to get more points!Here are final standings, frankly I'm quite impressed by Cevdet books!E.g approx.80-90 Elo better than Perfect 2021 book... just unbelievable!That means also, in overall openings Cevdet books are superior strong!!Plus, Cerebellum, Katzenmaier, DON, TrendMoves etc performed very well too!
CS LI's Champion: Cevdet2-BIN - Congratulations to Cevdet Sarı!
# PLAYER RATING POINTS GAMES (%) 1 Stockfish 170121 : 3712.9 1004.5 1370 73.3% 2 Eman 6.91 : 3712.3 934.0 1250 74.7% 3 CF EXT 161120 : 3708.2 852.5 1170 72.9% 4 Fat Fritz 2 : 3675.4 799.0 1150 69.5% 5 Komodo Dragon : 3650.0 896.5 1370 65.4% 6 Cevdet2-BIN HD : 3563.8 154.5 300 51.5% 7 Perfect2 Cevdet HD : 3558.3 152.5 300 50.8% 8 Cerebellum 3M Dual HD : 3555.5 151.5 300 50.5% 9 Katzenmaier1701 HD : 3554.1 151.0 300 50.3% 10 DON Lite 3 HD : 3543.3 147.0 300 49.0% 11 TrendMoves2020 HD : 3541.5 146.5 300 48.8% 12 1337chess 041020 HD : 3537.3 145.0 300 48.3% 13 Cevdet2-BIN KO : 3536.0 144.5 300 48.2% 14 RadaR 050121 HD : 3536.0 144.5 300 48.2% 15 M11.3 HD : 3536.0 144.5 300 48.2% 16 Alchemist 110121 HD : 3531.8 143.0 300 47.7% 17 HurnaTal v2 HD : 3523.4 140.0 300 46.7% 18 2600-Plus-Club HD : 3523.4 140.0 300 46.7% 19 TR Power HD : 3515.0 137.0 300 45.7% 20 NimaTiv Beta HD : 3509.3 135.0 300 45.0% 21 Perfect 2021 HD : 3472.3 406.5 1000 40.6% 22 SlowChess 2.5 : 3458.8 565.5 1450 39.0% 23 Nemorino 6.00 : 3453.9 556.5 1450 38.4% 24 RubiChess 2.0.1 : 3396.5 452.0 1450 31.2% 25 Minic 2.53 : 3370.3 393.5 1350 29.1% 26 Igel 2.9.0 : 3370.1 393.0 1350 29.1%
• Calculated by Ordo chess program / KO = Komodo 13.2.5 / HD = Houdini 6.03
More Details: • Red Name = Deep books, played by same chess engine: Houdini 6.03 Exception, The Champion (Cevdet2) is tested via Komodo 13.2.5 too • Black Name = Engines are used to play via Perfect 2021 opening book • As far as possible, as book participants are played Top ranked of CS L • All opening books are used to play by Author's recommended settings • New book entries are played min 300 games per player (vs NNUE...) • The overall draw percentage (based on 9430 games) not so high: 53% • Download all of the played tour games (in PGN format): SCCT CS LI
50 49 48 47 46 45 44 43 42 41 40 39 38 37 36 35 34 33 32 31 30