Can Komodo give a Queen to a 1500 chess Engine on CCRL?

[Chessqueen]

Dragon2 using standard Dragon2 NOT MCTS 1 thread or core (Contempt = 100) Intel i7 3.40GHz, to use 1 core is much better than using all my 4 cores with Knight Odds, since according to Mr. Kaufman the reason one thread is enough is that once Dragon2 outsearch his opponent engine by something like ten plies, searching even deeper will just lead Dragon2 to avoid things that his opponent engine would be unlikely to find in any case.

[pgn][Event "Queen Odds"]
[Date "2021.09.09"]
[Round "?"]
[White "Dragon-2-64bit-avx2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "21:29:34"]
[WhiteElo "3570"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "257"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Qa8+ Kd7 18. Rfd1+ Ke8 19. Rxd8+
Nxd8 20. Rd1 Nc6 21. Qc8 Ne5 22. hxg4 Nd3 23. Qxc7 Qh4 24. f3 Nxc5 25. Qxc5
Nc6 26. Qb5 Qh2+ 27. Kf1 Qc7 28. Qxc4 Rh2 29. Kf2 Qb6+ 30. Kg3 Rh6 31. Qa4
Qc7+ 32. f4 f6 33. Qb5 e5 34. f5 Bf7 35. Bf3 Bxa2 36. c4 Qa5 37. Qxb7 Ne7
38. Qd7+ Kf8 39. Qb5 Qc7 40. c5 Nc6 41. Rd6 Nd4 42. Qb6 Qxb6 43. cxb6 Ke7
44. b7 Rh8 45. Ra6 Rb8 46. Rxa2 Rxb7 47. Bg2 Rc7 48. g5 fxg5 49. Kg4 Kf6
50. Ra1 g6 51. fxg6 Kxg6 52. Ra5 Kf6 53. Bh3 Rc3 54. Ra6+ Ne6 55. Rb6 Ra3
56. Bf1 Re3 57. Bg2 Rd3 58. Bf3 Rc3 59. Bd1 Ke7 60. Kf5 Nc5 61. Rh6 Re3 62.
Rc6 Nxe4 63. Re6+ Kf8 64. Kg6 Nc5 65. Rc6 Rc3 66. Rc7 Ke8 67. Ra7 Rc4 68.
Kxg5 Rf4 69. Kh5 Rf2 70. Ra5 Nd3 71. Ra7 Kd8 72. Bg4 Nc5 73. Kh4 e4 74. Kh3
Rf1 75. Kg2 Rf4 76. Kg3 Rf6 77. Ra5 Rc6 78. Kf4 Kc7 79. Ke5 Nb7 80. Ra2 Nd6
81. Kd5 Nb5 82. Rb2 Nc3+ 83. Kd4 Kd6 84. Rb7 Ra6 85. Kxc3 Ke5 86. Re7+ Kf4
87. Be6 Ra3+ 88. Kc4 Ra1 89. Bd5 e3 90. Rf7+ Kg5 91. Bf3 Ra6 92. Kd4 Re6
93. Be2 Kh6 94. Rf3 Kg5 95. Rf7 Kh4 96. Rg7 Re8 97. Kd3 Kh3 98. Bf3 Kh2 99.
Ke2 Re5 100. Bg2 Re8 101. Rg5 Re7 102. Rg6 Re5 103. Rg8 Re7 104. Rg5 Re8
105. Rg7 Re5 106. Rg6 Ra5 107. Kf3 Rh5 108. Kxe3 Ra5 109. Kf4 Rh5 110. Be4
Rh4+ 111. Ke3 Rh3+ 112. Bf3 Rg3 113. Rh6+ Rh3 114. Rd6 Rg3 115. Kf4 Rg8
116. Rh6+ Kg1 117. Ke3 Kf1 118. Rb6 Re8+ 119. Be4 Kg1 120. Rh6 Rxe4+ 121.
Kxe4 Kf2 122. Kf4 Ke2 123. Rd6 Kf2 124. Rd2+ Ke1 125. Ke3 Kf1 126. Rb2 Kg1
127. Kf3 Kh1 128. Kg3 Kg1 129. Rb1# 1-0[/pgn]

[lkaufman]

Dragon2 using standard Dragon2 NOT MCTS 1 thread or core (Contempt = 100) Intel i7 3.40GHz, to use 1 core is much better than using all my 4 cores with Knight Odds, since according to Mr. Kaufman the reason one thread is enough is that once Dragon2 outsearch his opponent engine by something like ten plies, searching even deeper will just lead Dragon2 to avoid things that his opponent engine would be unlikely to find in any case.

[pgn][Event "Queen Odds"]
[Date "2021.09.09"]
[Round "?"]
[White "Dragon-2-64bit-avx2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "21:29:34"]
[WhiteElo "3570"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "257"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Qa8+ Kd7 18. Rfd1+ Ke8 19. Rxd8+
Nxd8 20. Rd1 Nc6 21. Qc8 Ne5 22. hxg4 Nd3 23. Qxc7 Qh4 24. f3 Nxc5 25. Qxc5
Nc6 26. Qb5 Qh2+ 27. Kf1 Qc7 28. Qxc4 Rh2 29. Kf2 Qb6+ 30. Kg3 Rh6 31. Qa4
Qc7+ 32. f4 f6 33. Qb5 e5 34. f5 Bf7 35. Bf3 Bxa2 36. c4 Qa5 37. Qxb7 Ne7
38. Qd7+ Kf8 39. Qb5 Qc7 40. c5 Nc6 41. Rd6 Nd4 42. Qb6 Qxb6 43. cxb6 Ke7
44. b7 Rh8 45. Ra6 Rb8 46. Rxa2 Rxb7 47. Bg2 Rc7 48. g5 fxg5 49. Kg4 Kf6
50. Ra1 g6 51. fxg6 Kxg6 52. Ra5 Kf6 53. Bh3 Rc3 54. Ra6+ Ne6 55. Rb6 Ra3
56. Bf1 Re3 57. Bg2 Rd3 58. Bf3 Rc3 59. Bd1 Ke7 60. Kf5 Nc5 61. Rh6 Re3 62.
Rc6 Nxe4 63. Re6+ Kf8 64. Kg6 Nc5 65. Rc6 Rc3 66. Rc7 Ke8 67. Ra7 Rc4 68.
Kxg5 Rf4 69. Kh5 Rf2 70. Ra5 Nd3 71. Ra7 Kd8 72. Bg4 Nc5 73. Kh4 e4 74. Kh3
Rf1 75. Kg2 Rf4 76. Kg3 Rf6 77. Ra5 Rc6 78. Kf4 Kc7 79. Ke5 Nb7 80. Ra2 Nd6
81. Kd5 Nb5 82. Rb2 Nc3+ 83. Kd4 Kd6 84. Rb7 Ra6 85. Kxc3 Ke5 86. Re7+ Kf4
87. Be6 Ra3+ 88. Kc4 Ra1 89. Bd5 e3 90. Rf7+ Kg5 91. Bf3 Ra6 92. Kd4 Re6
93. Be2 Kh6 94. Rf3 Kg5 95. Rf7 Kh4 96. Rg7 Re8 97. Kd3 Kh3 98. Bf3 Kh2 99.
Ke2 Re5 100. Bg2 Re8 101. Rg5 Re7 102. Rg6 Re5 103. Rg8 Re7 104. Rg5 Re8
105. Rg7 Re5 106. Rg6 Ra5 107. Kf3 Rh5 108. Kxe3 Ra5 109. Kf4 Rh5 110. Be4
Rh4+ 111. Ke3 Rh3+ 112. Bf3 Rg3 113. Rh6+ Rh3 114. Rd6 Rg3 115. Kf4 Rg8
116. Rh6+ Kg1 117. Ke3 Kf1 118. Rb6 Re8+ 119. Be4 Kg1 120. Rh6 Rxe4+ 121.
Kxe4 Kf2 122. Kf4 Ke2 123. Rd6 Kf2 124. Rd2+ Ke1 125. Ke3 Kf1 126. Rb2 Kg1
127. Kf3 Kh1 128. Kg3 Kg1 129. Rb1# 1-0[/pgn]

Yes indeed, that's what I've been saying, although you mistakenly labelled this game "queen odds" when you meant "knight odds"; "Benjamin" is probably stronger than Magnus Carlsen in Rapid or Blitz and especially good at winning with a handicap, so for any program to defeat it at knight odds is something. Now we need to see if this applies against humans; see if you can get your friend Jorge Hasbun to play it at knight odds in Rapid using just one thread. I think you can even give him substantial time odds as well; it just doesn't seem to be helpful to go beyond about 22 plies when giving knight odds.

[Chessqueen]

Dragon2 using standard Dragon2 NOT MCTS 1 thread or core (Contempt = 100) Intel i7 3.40GHz, to use 1 core is much better than using all my 4 cores with Knight Odds, since according to Mr. Kaufman the reason one thread is enough is that once Dragon2 outsearch his opponent engine by something like ten plies, searching even deeper will just lead Dragon2 to avoid things that his opponent engine would be unlikely to find in any case.

[pgn][Event "Knight Odds"]
[Date "2021.09.09"]
[Round "?"]
[White "Dragon-2-64bit-avx2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "21:29:34"]
[WhiteElo "3570"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "257"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Qa8+ Kd7 18. Rfd1+ Ke8 19. Rxd8+
Nxd8 20. Rd1 Nc6 21. Qc8 Ne5 22. hxg4 Nd3 23. Qxc7 Qh4 24. f3 Nxc5 25. Qxc5
Nc6 26. Qb5 Qh2+ 27. Kf1 Qc7 28. Qxc4 Rh2 29. Kf2 Qb6+ 30. Kg3 Rh6 31. Qa4
Qc7+ 32. f4 f6 33. Qb5 e5 34. f5 Bf7 35. Bf3 Bxa2 36. c4 Qa5 37. Qxb7 Ne7
38. Qd7+ Kf8 39. Qb5 Qc7 40. c5 Nc6 41. Rd6 Nd4 42. Qb6 Qxb6 43. cxb6 Ke7
44. b7 Rh8 45. Ra6 Rb8 46. Rxa2 Rxb7 47. Bg2 Rc7 48. g5 fxg5 49. Kg4 Kf6
50. Ra1 g6 51. fxg6 Kxg6 52. Ra5 Kf6 53. Bh3 Rc3 54. Ra6+ Ne6 55. Rb6 Ra3
56. Bf1 Re3 57. Bg2 Rd3 58. Bf3 Rc3 59. Bd1 Ke7 60. Kf5 Nc5 61. Rh6 Re3 62.
Rc6 Nxe4 63. Re6+ Kf8 64. Kg6 Nc5 65. Rc6 Rc3 66. Rc7 Ke8 67. Ra7 Rc4 68.
Kxg5 Rf4 69. Kh5 Rf2 70. Ra5 Nd3 71. Ra7 Kd8 72. Bg4 Nc5 73. Kh4 e4 74. Kh3
Rf1 75. Kg2 Rf4 76. Kg3 Rf6 77. Ra5 Rc6 78. Kf4 Kc7 79. Ke5 Nb7 80. Ra2 Nd6
81. Kd5 Nb5 82. Rb2 Nc3+ 83. Kd4 Kd6 84. Rb7 Ra6 85. Kxc3 Ke5 86. Re7+ Kf4
87. Be6 Ra3+ 88. Kc4 Ra1 89. Bd5 e3 90. Rf7+ Kg5 91. Bf3 Ra6 92. Kd4 Re6
93. Be2 Kh6 94. Rf3 Kg5 95. Rf7 Kh4 96. Rg7 Re8 97. Kd3 Kh3 98. Bf3 Kh2 99.
Ke2 Re5 100. Bg2 Re8 101. Rg5 Re7 102. Rg6 Re5 103. Rg8 Re7 104. Rg5 Re8
105. Rg7 Re5 106. Rg6 Ra5 107. Kf3 Rh5 108. Kxe3 Ra5 109. Kf4 Rh5 110. Be4
Rh4+ 111. Ke3 Rh3+ 112. Bf3 Rg3 113. Rh6+ Rh3 114. Rd6 Rg3 115. Kf4 Rg8
116. Rh6+ Kg1 117. Ke3 Kf1 118. Rb6 Re8+ 119. Be4 Kg1 120. Rh6 Rxe4+ 121.
Kxe4 Kf2 122. Kf4 Ke2 123. Rd6 Kf2 124. Rd2+ Ke1 125. Ke3 Kf1 126. Rb2 Kg1
127. Kf3 Kh1 128. Kg3 Kg1 129. Rb1# 1-0[/pgn]

Yes indeed, that's what I've been saying, although you mistakenly labelled this game "queen odds" when you meant "knight odds"; "Benjamin" is probably stronger than Magnus Carlsen in Rapid or Blitz and especially good at winning with a handicap, so for any program to defeat it at knight odds is something. Now we need to see if this applies against humans; see if you can get your friend Jorge Hasbun to play it at knight odds in Rapid using just one thread. I think you can even give him substantial time odds as well; it just doesn't seem to be helpful to go beyond about 22 plies when giving knight odds.

I believe that for the same reason it is not advisable to run Stockfish with more than 1 core

[Chessqueen]

I believe that for the same reason it is not advisable to run Stockfish with more than 1 core, but Komodo Dragon still has the advantage of being able to set different Contempt depending on what Odds is being given, for instance with no MCTS I use Contempt = 100, I still do not Know what Contempt to use for Rook Odds, or should I use 140 ? and for Queen 200

Note: Wish me Luck with Dragon2 I agreed to play 2 games Vs Jose Lisandro Munoz Santana at TC 30'+10" Knight Odds starting in 5 Minutes via Skype and two other players will be by his side including another of my friend, which I see via Skype. I agree to pay him $50.00 for a win and $25.00 for a draw, he is believing that he will have an easy 2 games not knowing that Komodo Dragon2 on my intel i7 3.40 GHz at Knight Odds is about 2600. He is thinking that he has an easy pay day since in Dominican Republic he makes only around $30.00 per day and he is well educated. This is the person who will play he is around 2420 FIDE https://ratings.fide.com/profile/6400361/calculations. Mr. Kaufman should contact GM Nihal and offer him a 3 games match at Knight Odds TC 30'+10" https://ratings.fide.com/profile/25092340

[pgn][Event "Knight Odds"]
[Date "2021.10.09"]
[Round "?"]
[White "Stockfish_14_x64_bmi2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "09:29:34"]
[WhiteElo "3585"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "167"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Bxe7 Qxe7 18. Qa8+ Kd7 19. Qxb7 Qe8
20. Rbd1+ Ke7 21. Qxc7+ Kf8 22. Qxc6 gxh3 23. Qb6 Ra8 24. Bf3 Rh4 25. Kh2
e5 26. Rd5 Rf4 27. Rfd1 Rf6 28. Qc7 Rc6 29. Qxe5 Qxe5+ 30. Rxe5 Rca6 31.
Rc5 Rxa2 32. Kg3 h2 33. Rxc4 R2a4 34. Rxa4 Rxa4 35. Re1 Ra2 36. Bg2 Ke7 37.
f3 Kd6 38. Kxh2 Kc5 39. Kg3 Rc2 40. Re3 Rc1 41. f4 Kc4 42. f5 Bh7 43. Kf4
Rxc3 44. Re1 g6 45. Bf1+ Kd4 46. Rd1+ Kc5 47. Rd5+ Kb6 48. f6 Rc1 49. Be2
Rc2 50. Bg4 Rc4 51. Bf3 Rc3 52. Re5 g5+ 53. Rxg5 Rc6 54. e5 Rc4+ 55. Kg3
Rc5 56. Bg4 Rc3+ 57. Kh4 Rc4 58. Rg7 Bf5 59. Kh5 Bxg4+ 60. Rxg4 Rc1 61. Kh6
Kc6 62. Kg7 Kd5 63. Kxf7 Rc7+ 64. Kg8 Ke6 65. Rg6 Rc8+ 66. Kg7 Kxe5 67. f7
Ke4 68. f8=Q Rxf8 69. Kxf8 Kf5 70. Rd6 Ke4 71. Kf7 Ke5 72. Rd1 Kf4 73. Kf6
Ke4 74. Rd6 Kf4 75. Re6 Kg4 76. Re4+ Kg3 77. Kg5 Kf3 78. Re7 Kg3 79. Rf7
Kh2 80. Kg4 Kg2 81. Rf6 Kh1 82. Kg3 Kg1 83. Rf8 Kh1 84. Rf1# *[/pgn]

[lkaufman]

I believe that for the same reason it is not advisable to run Stockfish with more than 1 core, but Komodo Dragon still has the advantage of being able to set different Contempt depending on what Odds is being given, for instance with no MCTS I use Contempt = 100, I still do not Know what Contempt to use for Rook Odds, or should I use 140 ? and for Queen 200

Note: Wish me Luck with Dragon2 I agreed to play 2 games Vs Jose Lisandro Munoz Santana at TC 30'+10" Knight Odds starting in 5 Minutes via Skype and two other players will be by his side including another of my friend, which I see via Skype. I agree to pay him $50.00 for a win and $25.00 for a draw, he is believing that he will have an easy 2 games not knowing that Komodo Dragon2 on my intel i7 3.40 GHz at Knight Odds is about 2600. He is thinking that he has an easy pay day since in Dominican Republic he makes only around $30.00 per day and he is well educated. This is the person who will play he is around 2420 FIDE https://ratings.fide.com/profile/6400361/calculations. Mr. Kaufman should contact GM Nihal and offer him a 3 games match at Knight Odds TC 30'+10" https://ratings.fide.com/profile/25092340

[pgn][Event "Knight Odds"]
[Date "2021.10.09"]
[Round "?"]
[White "Stockfish_14_x64_bmi2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "09:29:34"]
[WhiteElo "3585"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "167"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Bxe7 Qxe7 18. Qa8+ Kd7 19. Qxb7 Qe8
20. Rbd1+ Ke7 21. Qxc7+ Kf8 22. Qxc6 gxh3 23. Qb6 Ra8 24. Bf3 Rh4 25. Kh2
e5 26. Rd5 Rf4 27. Rfd1 Rf6 28. Qc7 Rc6 29. Qxe5 Qxe5+ 30. Rxe5 Rca6 31.
Rc5 Rxa2 32. Kg3 h2 33. Rxc4 R2a4 34. Rxa4 Rxa4 35. Re1 Ra2 36. Bg2 Ke7 37.
f3 Kd6 38. Kxh2 Kc5 39. Kg3 Rc2 40. Re3 Rc1 41. f4 Kc4 42. f5 Bh7 43. Kf4
Rxc3 44. Re1 g6 45. Bf1+ Kd4 46. Rd1+ Kc5 47. Rd5+ Kb6 48. f6 Rc1 49. Be2
Rc2 50. Bg4 Rc4 51. Bf3 Rc3 52. Re5 g5+ 53. Rxg5 Rc6 54. e5 Rc4+ 55. Kg3
Rc5 56. Bg4 Rc3+ 57. Kh4 Rc4 58. Rg7 Bf5 59. Kh5 Bxg4+ 60. Rxg4 Rc1 61. Kh6
Kc6 62. Kg7 Kd5 63. Kxf7 Rc7+ 64. Kg8 Ke6 65. Rg6 Rc8+ 66. Kg7 Kxe5 67. f7
Ke4 68. f8=Q Rxf8 69. Kxf8 Kf5 70. Rd6 Ke4 71. Kf7 Ke5 72. Rd1 Kf4 73. Kf6
Ke4 74. Rd6 Kf4 75. Re6 Kg4 76. Re4+ Kg3 77. Kg5 Kf3 78. Re7 Kg3 79. Rf7
Kh2 80. Kg4 Kg2 81. Rf6 Kh1 82. Kg3 Kg1 83. Rf8 Kh1 84. Rf1# *[/pgn]

Remember, the performance rating of Komodo Dragon 2 giving knight odds to HUMAN players at 15' + 10" is 2460 FIDE in fifteen games; it would presumably be somewhat lower, maybe around 2420, at 30' + 10". It's true that Dragon used MCTS with many threads in those games, while you are presumably using no MCTS and single thread, but it's not yet clear which approach is better vs humans. So I would rate your match as a tossup. You could perhaps improve your results by using all four threads but limiting the search to 22 plies, assuming your GUI will do that; this will allow Dragon to move much faster, depriving the opponent of Ponder time, a major factor for humans I think.
Regarding Contempt, I use 100 for knight odds vs engines, 125 vs humans, 25 more for rook odds, 200 for queen odds. It doesn't matter much if you use slightly different numbers, I don't really know what is best vs. humans, I can only get huge numbers of games vs. engines.

[Chessqueen]

I believe that for the same reason it is not advisable to run Stockfish with more than 1 core, but Komodo Dragon still has the advantage of being able to set different Contempt depending on what Odds is being given, for instance with no MCTS I use Contempt = 100, I still do not Know what Contempt to use for Rook Odds, or should I use 140 ? and for Queen 200

Note: Wish me Luck with Dragon2 I agreed to play 2 games Vs Jose Lisandro Munoz Santana at TC 30'+10" Knight Odds starting in 5 Minutes via Skype and two other players will be by his side including another of my friend, which I see via Skype. I agree to pay him $50.00 for a win and $25.00 for a draw, he is believing that he will have an easy 2 games not knowing that Komodo Dragon2 on my intel i7 3.40 GHz at Knight Odds is about 2600. He is thinking that he has an easy pay day since in Dominican Republic he makes only around $30.00 per day and he is well educated. This is the person who will play he is around 2420 FIDE https://ratings.fide.com/profile/6400361/calculations. Mr. Kaufman should contact GM Nihal and offer him a 3 games match at Knight Odds TC 30'+10" https://ratings.fide.com/profile/25092340

[pgn][Event "Knight Odds"]
[Date "2021.10.09"]
[Round "?"]
[White "Stockfish_14_x64_bmi2"]
[Black "Benjamin"]
[Result "1-0"]
[BlackElo "2600"]
[Time "09:29:34"]
[WhiteElo "3585"]
[TimeControl "300+1"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNBQKB1R w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "167"]
[WhiteType "program"]
[BlackType "program"]

1. d4 d5 2. Bf4 Bf5 3. c4 e6 4. Nc3 Bb4 5. g3 dxc4 6. Qa4+ Nc6 7. Bg2 Ne7
8. e4 Bg4 9. h3 Bh5 10. g4 Qxd4 11. O-O Qf6 12. Be3 Bxc3 13. bxc3 Bg6 14.
Rab1 O-O-O 15. Bxa7 h5 16. Bc5 hxg4 17. Bxe7 Qxe7 18. Qa8+ Kd7 19. Qxb7 Qe8
20. Rbd1+ Ke7 21. Qxc7+ Kf8 22. Qxc6 gxh3 23. Qb6 Ra8 24. Bf3 Rh4 25. Kh2
e5 26. Rd5 Rf4 27. Rfd1 Rf6 28. Qc7 Rc6 29. Qxe5 Qxe5+ 30. Rxe5 Rca6 31.
Rc5 Rxa2 32. Kg3 h2 33. Rxc4 R2a4 34. Rxa4 Rxa4 35. Re1 Ra2 36. Bg2 Ke7 37.
f3 Kd6 38. Kxh2 Kc5 39. Kg3 Rc2 40. Re3 Rc1 41. f4 Kc4 42. f5 Bh7 43. Kf4
Rxc3 44. Re1 g6 45. Bf1+ Kd4 46. Rd1+ Kc5 47. Rd5+ Kb6 48. f6 Rc1 49. Be2
Rc2 50. Bg4 Rc4 51. Bf3 Rc3 52. Re5 g5+ 53. Rxg5 Rc6 54. e5 Rc4+ 55. Kg3
Rc5 56. Bg4 Rc3+ 57. Kh4 Rc4 58. Rg7 Bf5 59. Kh5 Bxg4+ 60. Rxg4 Rc1 61. Kh6
Kc6 62. Kg7 Kd5 63. Kxf7 Rc7+ 64. Kg8 Ke6 65. Rg6 Rc8+ 66. Kg7 Kxe5 67. f7
Ke4 68. f8=Q Rxf8 69. Kxf8 Kf5 70. Rd6 Ke4 71. Kf7 Ke5 72. Rd1 Kf4 73. Kf6
Ke4 74. Rd6 Kf4 75. Re6 Kg4 76. Re4+ Kg3 77. Kg5 Kf3 78. Re7 Kg3 79. Rf7
Kh2 80. Kg4 Kg2 81. Rf6 Kh1 82. Kg3 Kg1 83. Rf8 Kh1 84. Rf1# *[/pgn]

Remember, the performance rating of Komodo Dragon 2 giving knight odds to HUMAN players at 15' + 10" is 2460 FIDE in fifteen games; it would presumably be somewhat lower, maybe around 2420, at 30' + 10". It's true that Dragon used MCTS with many threads in those games, while you are presumably using no MCTS and single thread, but it's not yet clear which approach is better vs humans. So I would rate your match as a tossup. You could perhaps improve your results by using all four threads but limiting the search to 22 plies, assuming your GUI will do that; this will allow Dragon to move much faster, depriving the opponent of Ponder time, a major factor for humans I think.

Well I screw up, but it is only $50.00 this is the game is progress via Skype
http://talkchess.com/forum3/viewtopic.php?f=2&t=78147

[bob]

But against another 1600 rated player Komodo lost very quickly, and I lost $25.00. This is using my Laptop Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory

[pgn][Event "Queen Odds"]
[Site "Philippines Chess Club"]
[Date "2021.09.05"]
[Round "2"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Jose"]
[Result "0-1"]
[BlackElo "1600"]
[Time "08:09:17"]
[WhiteElo "3470"]
[TimeControl "300+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNB1KBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "43"]
[WhiteType "program"]
[BlackType "human"]

1. d4 Nf6 2. Bf4 d5 3. Nf3 Nc6 4. e3 Bg4 5. Nbd2 Bxf3 6. gxf3 Nh5 7. Bg3
Nxg3 8. hxg3 e5 9. O-O-O exd4 10. exd4 Nxd4 11. Re1+ Be7 12. c3 Ne6 13. Re5
O-O 14. f4 Bf6 15. Reh5 h6 16. Rxh6 gxh6 17. Kd1 Bg7 18. f3 c5 19. Kc2 d4
20. Bc4 dxc3 21. bxc3 {White resigns} Qd6 22. Ne4 {White resigns} *[/pgn]

This is strange because the blunder 16. Rxh6?? makes no sense, I can't replicate it at any time or depth with K14.1 (I don't have 13.3 on my laptop). Were you perhaps using MCTS? I think MCTS just falls apart completely when down a huge amount like this. I would only use MCTS for handicap play if the handicap is not more than a rook, if the time control is Rapid rather than blitz, and if the number of threads is at least four.

You might investigate how the rating was forced that low? Doesn't take a whole lot of randomnesss to introduce a gross blunder when the randomness is given a high probability to produce really low ratings.

I think you misunderstand something here. White is full strength Komodo giving queen odds; Black is a human. There is no forced low rating or randomness involved. The explanation turned out to be that Komodo was in MCTS mode, which works fine when down a piece or so, but when down a queen or more it is apt to make crazy sacrifices on the theory that one chance in a million is better than none at all!

I was more thinking about the weakened opponents. And how they do it. I know what I do in Crafty and it is REALLY quite hard to make it play like a REAL 1600 player, for example. 1600 players don't play lots of really good moves followed by a blunder. I used randomness to help with this, but even though the Elo numbers are reasonably accurate in computer vs computer matches, it still doesn't "feel right" to me.

[amanjpro]

But against another 1600 rated player Komodo lost very quickly, and I lost $25.00. This is using my Laptop Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory

[pgn][Event "Queen Odds"]
[Site "Philippines Chess Club"]
[Date "2021.09.05"]
[Round "2"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Jose"]
[Result "0-1"]
[BlackElo "1600"]
[Time "08:09:17"]
[WhiteElo "3470"]
[TimeControl "300+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNB1KBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "43"]
[WhiteType "program"]
[BlackType "human"]

1. d4 Nf6 2. Bf4 d5 3. Nf3 Nc6 4. e3 Bg4 5. Nbd2 Bxf3 6. gxf3 Nh5 7. Bg3
Nxg3 8. hxg3 e5 9. O-O-O exd4 10. exd4 Nxd4 11. Re1+ Be7 12. c3 Ne6 13. Re5
O-O 14. f4 Bf6 15. Reh5 h6 16. Rxh6 gxh6 17. Kd1 Bg7 18. f3 c5 19. Kc2 d4
20. Bc4 dxc3 21. bxc3 {White resigns} Qd6 22. Ne4 {White resigns} *[/pgn]

This is strange because the blunder 16. Rxh6?? makes no sense, I can't replicate it at any time or depth with K14.1 (I don't have 13.3 on my laptop). Were you perhaps using MCTS? I think MCTS just falls apart completely when down a huge amount like this. I would only use MCTS for handicap play if the handicap is not more than a rook, if the time control is Rapid rather than blitz, and if the number of threads is at least four.

You might investigate how the rating was forced that low? Doesn't take a whole lot of randomnesss to introduce a gross blunder when the randomness is given a high probability to produce really low ratings.

I think you misunderstand something here. White is full strength Komodo giving queen odds; Black is a human. There is no forced low rating or randomness involved. The explanation turned out to be that Komodo was in MCTS mode, which works fine when down a piece or so, but when down a queen or more it is apt to make crazy sacrifices on the theory that one chance in a million is better than none at all!

I was more thinking about the weakened opponents. And how they do it. I know what I do in Crafty and it is REALLY quite hard to make it play like a REAL 1600 player, for example. 1600 players don't play lots of really good moves followed by a blunder. I used randomness to help with this, but even though the Elo numbers are reasonably accurate in computer vs computer matches, it still doesn't "feel right" to me.

With the surge of NNUE evals, would it make sense to train a few nets, with different strengths, and include them all in the binary,, each to be used for a level... That is what I'm going to try next in my engine

[lkaufman]

But against another 1600 rated player Komodo lost very quickly, and I lost $25.00. This is using my Laptop Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory

[pgn][Event "Queen Odds"]
[Site "Philippines Chess Club"]
[Date "2021.09.05"]
[Round "2"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Jose"]
[Result "0-1"]
[BlackElo "1600"]
[Time "08:09:17"]
[WhiteElo "3470"]
[TimeControl "300+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNB1KBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "43"]
[WhiteType "program"]
[BlackType "human"]

1. d4 Nf6 2. Bf4 d5 3. Nf3 Nc6 4. e3 Bg4 5. Nbd2 Bxf3 6. gxf3 Nh5 7. Bg3
Nxg3 8. hxg3 e5 9. O-O-O exd4 10. exd4 Nxd4 11. Re1+ Be7 12. c3 Ne6 13. Re5
O-O 14. f4 Bf6 15. Reh5 h6 16. Rxh6 gxh6 17. Kd1 Bg7 18. f3 c5 19. Kc2 d4
20. Bc4 dxc3 21. bxc3 {White resigns} Qd6 22. Ne4 {White resigns} *[/pgn]

This is strange because the blunder 16. Rxh6?? makes no sense, I can't replicate it at any time or depth with K14.1 (I don't have 13.3 on my laptop). Were you perhaps using MCTS? I think MCTS just falls apart completely when down a huge amount like this. I would only use MCTS for handicap play if the handicap is not more than a rook, if the time control is Rapid rather than blitz, and if the number of threads is at least four.

You might investigate how the rating was forced that low? Doesn't take a whole lot of randomnesss to introduce a gross blunder when the randomness is given a high probability to produce really low ratings.

I think you misunderstand something here. White is full strength Komodo giving queen odds; Black is a human. There is no forced low rating or randomness involved. The explanation turned out to be that Komodo was in MCTS mode, which works fine when down a piece or so, but when down a queen or more it is apt to make crazy sacrifices on the theory that one chance in a million is better than none at all!

I was more thinking about the weakened opponents. And how they do it. I know what I do in Crafty and it is REALLY quite hard to make it play like a REAL 1600 player, for example. 1600 players don't play lots of really good moves followed by a blunder. I used randomness to help with this, but even though the Elo numbers are reasonably accurate in computer vs computer matches, it still doesn't "feel right" to me.

With the surge of NNUE evals, would it make sense to train a few nets, with different strengths, and include them all in the binary,, each to be used for a level... That is what I'm going to try next in my engine

Yes, it makes sense, but search depth is also a key factor in simulating human play; 1600 rated players don't usually search 15 ply deep in choosing their moves, so neither should a simulation. Getting all this right is a very difficult problem.

[Angrim]

But against another 1600 rated player Komodo lost very quickly, and I lost $25.00. This is using my Laptop Intel(R) Core(TM) i7-4770 CPU @ 3.40GHz with 15.9 GB Memory

[pgn][Event "Queen Odds"]
[Site "Philippines Chess Club"]
[Date "2021.09.05"]
[Round "2"]
[White "Komodo-13.3-64bit-bmi2"]
[Black "Jose"]
[Result "0-1"]
[BlackElo "1600"]
[Time "08:09:17"]
[WhiteElo "3470"]
[TimeControl "300+3"]
[SetUp "1"]
[FEN "rnbqkbnr/pppppppp/8/8/8/8/PPPPPPPP/RNB1KBNR w KQkq - 0 1"]
[Termination "normal"]
[PlyCount "43"]
[WhiteType "program"]
[BlackType "human"]

1. d4 Nf6 2. Bf4 d5 3. Nf3 Nc6 4. e3 Bg4 5. Nbd2 Bxf3 6. gxf3 Nh5 7. Bg3
Nxg3 8. hxg3 e5 9. O-O-O exd4 10. exd4 Nxd4 11. Re1+ Be7 12. c3 Ne6 13. Re5
O-O 14. f4 Bf6 15. Reh5 h6 16. Rxh6 gxh6 17. Kd1 Bg7 18. f3 c5 19. Kc2 d4
20. Bc4 dxc3 21. bxc3 {White resigns} Qd6 22. Ne4 {White resigns} *[/pgn]

This is strange because the blunder 16. Rxh6?? makes no sense, I can't replicate it at any time or depth with K14.1 (I don't have 13.3 on my laptop). Were you perhaps using MCTS? I think MCTS just falls apart completely when down a huge amount like this. I would only use MCTS for handicap play if the handicap is not more than a rook, if the time control is Rapid rather than blitz, and if the number of threads is at least four.

You might investigate how the rating was forced that low? Doesn't take a whole lot of randomnesss to introduce a gross blunder when the randomness is given a high probability to produce really low ratings.

I think you misunderstand something here. White is full strength Komodo giving queen odds; Black is a human. There is no forced low rating or randomness involved. The explanation turned out to be that Komodo was in MCTS mode, which works fine when down a piece or so, but when down a queen or more it is apt to make crazy sacrifices on the theory that one chance in a million is better than none at all!

I was more thinking about the weakened opponents. And how they do it. I know what I do in Crafty and it is REALLY quite hard to make it play like a REAL 1600 player, for example. 1600 players don't play lots of really good moves followed by a blunder. I used randomness to help with this, but even though the Elo numbers are reasonably accurate in computer vs computer matches, it still doesn't "feel right" to me.

With the surge of NNUE evals, would it make sense to train a few nets, with different strengths, and include them all in the binary,, each to be used for a level... That is what I'm going to try next in my engine

You might want to look at the maia project, they are already working on that.
https://maiachess.com/