Leela wins the match series against Komodo and wins a Pawn odds game against Stockfish!

Leela won 16-14 against Komodo the 30 games match to determine 3rd position in CCCC1, while on the variants tournament she won a game against Stockfish while being handicapped having a Pawn less!!

So the 3rd place Komodo-Leela match has ended with Leela winning 16-14, having 5 wins, 3 losses and 22 draws , so Leela had finished 3rd in the first CCCC tournament according to Chess.com changed rules.

One thing remains to be seen though if they will remove game 3 of this match where from a mistake/bug the same opening with before was repeated while it shouldn’t.
They had to use 15 predefined opening positions(for 3 first moves) with reversed colors, so each engine will have to play every position once with black and once with white.

So this normally happened in the first 2 games where Leela won both 2 games(as white and as black) with a specific opening(Torre Attack). Then while King’s Gambit opening should have been used in the 3rd game (Komodo had white), the previous Torre opening was used again. Komodo deviated early in move 10 so a completely different game has been played. Leela won again.
Thing is the final score is 16-14 and if we don’t count the 3rd game it would be 15-14, so Komodo would have a chance to tie the series with the Komodo- Leela game on King’s Gambit opening. Even though in that variation of King’s Gambit, engines favor black and usually engine games end with a draw or a black win but this has to be seen for the current pair.

Here are the 30 Leela-Komodo games : Leela-Komodo PGN

Meanwhile the Chess variants games have started.
First is the f2-Pawn missing variant where Leela already lost easily the games as white against Komodo and Ethereal, drew with black against Houdini and made a shocker win with white, against Stockfish!
Yes Leela being handicapped playing with white with a missing f2-Pawn , went on to win against Stockfish after slowly outplaying it to reach a won endgame. Stockfish made a mistake in a tough position very early but this was not the critical mistake since after that the position was equal anyway. Somehow without any apparent big mistake of black(Stockfish), black’s position got worse and worse and the endgame they reached was won for white, where Leela after her usual shuffling instead of immediate killing, managed to win.

It was also the first win with white pieces of any engine till now in this variant, after 9 black wins and 4 draws!!
For more info about it READ HERE where all 5 variants are explained.

Here is the Leela Stockfish game:

[Event “CCCC 1: Rapid Rumble (15|5) Bonus Games”] [Site “Chess.com”] [Date “2018.10.03”] [Round “?”] [White “Lc0 17.11089”] [Black “Stockfish 220818”] [Result “1-0”] [ECO “A02”] [WhiteElo “3300”] [BlackElo “3400”] [Annotator “Ts,George”] [PlyCount “235”] [EventDate “2018.??.??”] [EventType “rapid”] 1. f4 Nh6 2. f5 Nxf5 3. Nf3 Nh6 4. Ng1 Ng8 {This maneuver was made since this is a f-Pawn handicap game. Leela has a Pawn less and even more it is the critical f-Pawn for King safety. Yet somehow she managed to make Stockfish to lose….. Till that game 9 games had been played with this variant(f-Pawn missing for white) and we had 6 black wins and 3 draws. So Leela is the first engine that won with white pieces. And against which opponent? Stockfish!} 5. e4 e6 6. Nc3 d5 7. d4 Bb4 8. e5 Qh4+ $6 {Stockfish missed that Qe4+ is not working.} 9. g3 $1 {Stockfish expected Ke2 here. Leela knew better.} Qd8 (9… Qe4+ $2 10. Kf2 Qxh1 11. h3 $1 Bxc3 12. bxc3 Qe4 13. Nf3 {Queen is trapped.} Nc6 14. Bd3 f5 $16 {[%csl Re4]}) 10. a3 Bf8 11. Be3 Ne7 12. Nf3 Nf5 {Here Leela thinks this is a mistake. Now she went in a positive evaluaiton for her position. Stockfish said -0.57 in favor of black.} 13. Bg5 Be7 14. Bxe7 Nxe7 { Both thought now game is equal.} 15. Bd3 h6 16. b4 b6 17. O-O O-O {Leela after this move of Stockfish really likes her position.} 18. Ne2 c6 19. a4 c5 20. c3 Nbc6 21. Qd2 Bd7 22. Rae1 Rc8 $6 {It’s not that clear where black blundered or made a mistake and went to lose the game. But Stockfish step by step got into a losing position and yet it’s not clear what he should do differently in order to hold. Allowing b5 in next move is a bad idea anyway since white locks any counterplay.} 23. b5 {[%csl Gd2,Gd3,Ge1,Ge2,Ge5,Gf1,Gf3] This reminds the anticomputer play of the old times. closing the position and gathering your forces slowly in the side of the computer’s King ready to attack. This old method stopped working completely from about 2010-2011 where engines(software) and hardware advances, made it impossible to outsmart in this way the engines.} Na5 24. Nf4 cxd4 25. cxd4 Qc7 26. Rc1 Qd8 27. Rxc8 Qxc8 28. Nh5 Nc4 29. Qf4 Ng6 30. Nf6+ $3 Kh8 (30… gxf6 31. Qxh6 Ncxe5 (31… fxe5 32. Ng5 $18 {[%csl Rh7]} ) 32. dxe5 Qc5+ 33. Kg2 Qe7 34. h4 $18 {[%csl Rg6,Gh7][%cal Gh4h5,Gd3h7,Gh6h7]} ) 31. Qc1 Na5 32. Qd2 Nc4 33. Qc1 Na5 34. Bxg6 Qxc1 35. Rxc1 gxf6 36. Bh5 Nc4 37. exf6 Kg8 38. Ne5 Nxe5 39. dxe5 Rc8 40. Rxc8+ Bxc8 {Even though material is equal and white has no immediate threat, positionally this is winning for white as Leela has all the play and free movement with her King and black pieces are locked in cages and h-Pawn of black is dead.} 41. Be2 d4 (41… Bb7 {Trying to keep the d-Pawn is worthless.} 42. Bd3 {[%csl Rh6][%cal Gd3h7]} Kh8 43. Kf2 Kg8 44. Ke3 Kh8 45. Kf4 {[%csl Rh6] h-Pawn will fall}) 42. Bf3 h5 43. Bxh5 Bb7 44. Kf2 Be4 45. Ke1 d3 46. Kd2 Kf8 47. Ke3 Bh7 48. Bd1 Ke8 49. g4 Kd8 50. Kd2 Ke8 51. Ke3 Kf8 52. Kf3 Ke8 53. Kf2 Kd7 54. Ke3 Kc7 55. h4 Kc8 56. h5 Kd7 57. Kf3 Ke8 58. Kf2 Kf8 59. Ke3 Kg8 60. Kf4 Kh8 61. Kf3 Bg8 62. Ke4 Bh7+ 63. Kf4 Kg8 64. Ke3 Kf8 65. Kf2 Be4 66. Ke3 Bh7 67. Bb3 Kg8 68. Bc4 d2 69. Kxd2 Be4 70. Be2 Bb7 71. g5 Be4 72. Bd1 Kh8 73. Kc3 Kg8 74. Be2 Bf5 75. Kd2 Bb1 76. Bd1 Kf8 77. Kc3 Ke8 78. Kb2 Bh7 79. Kc3 Be4 80. Kd2 Bf5 81. Ke3 Bb1 82. Kf3 Kf8 83. Ke3 Ke8 84. Kd4 Kf8 85. Kc3 Bh7 86. Be2 Kg8 87. Bg4 Be4 88. Bd1 Kh8 89. Kd2 Bh7 90. Bg4 Kg8 91. Bd1 Kf8 92. Bb3 Ke8 93. Bc4 Kd7 94. Be2 Ke8 95. Kc3 Kf8 96. Bd1 Bb1 97. Bg4 Be4 98. Kd2 Kg8 99. g6 fxg6 100. Bxe6+ Kf8 101. h6 g5 102. Ke3 Bg6 103. Bd5 g4 104. Kf4 g3 105. Kxg3 Ke8 106. e6 Kd8 107. Kf4 a6 108. bxa6 Be8 109. a7 Kc7 110. a8=Q Kd6 111. h7 Kc5 112. h8=R Bd7 113. Ke5 Kb4 114. Rh3 Bxa4 115. Kd4 Bd1 116. Bc6 Bb3 117. f7 b5 118. f8=Q# 1-0

Testing about the aforementioned, in the above blog post, disparity in results/behavior of Leela between a FEN position and a PGN position(with history of moves that lead to the desired position), the following 2 gaunlets had Leela with Lc0v18 11089 +3-6 TBs versus: Stockfish dev, Stockfish 8, Fizbo 2, Ethereal 11, Andscacs 0.94, Booot 6.3.1 and Fire 7.1 from the “Up one row” position variant that they will play in the variants tournament.
Time control was 4’+2" and the Stockfish/Leela nodes per second ratio of 850 , the same Deepmind used for AlphaZero-Stockfish 8 games(while in CCCC they have 2000 , which, compared to 850, is more in favor for traditional engines obviously).

The “up one row” position:

The one gaunlet is with FEN of the variant:
rnbqkbnr/pppppppp/8/8/8/PPPPPPPP/RNBQKBNR/8 w kq - 0 1

The other gaunlet is with PGN of the variant:
[Event “?”]
[Site “?”]
[Date “????.??.??”]
[Round “?”]
[White “?”]
[Black “?”]
[Result “*”]

1. g3 Nf6 2. Bh3 Ng8 3. c3 Nf6 4. Bf5 Ng8 5. Bc2 Nf6 6. d3 Ng8 7. Be3 Nf6 8. f3
Ng8 9. Bf2 Nf6 10. e3 Ng8 11. Ne2 Nf6 12. Nf4 Ng8 13. Ng2 Nf6 14. Nd2 Ng8 15.
Nc4 Nf6 16. b3 Ng8 17. Nb2 Nf6 18. Ke2 Ng8 19. Qd2 Nf6 20. a3 Ng8 21. h3 Nf6 22.
Rh2 Ng8 23. Rc1 Nf6 24. Rb1 Ng8 25. Ra1 Nf6 26. Ra2 Ng8 *

And the results were:

The gaunlet with history/PGN:
Leela had 10.0/14

Lc0v18 11089     - Andscacs 0.94                     2.0 - 0.0    +2/=0/-0  

100.00%
Lc0v18 11089 - Booot 6.3.1 2.0 - 0.0 +2/=0/-0
100.00%
Lc0v18 11089 - Ethereal 11.00-x64-pext 1.0 - 1.0 +1/=0/-1
50.00%
Lc0v18 11089 - Fire 7.1 2.0 - 0.0 +2/=0/-0
100.00%
Lc0v18 11089 - Fizbo2x64_bmi2 1.5 - 0.5 +1/=1/-0
75.00%
Lc0v18 11089 - Stockfish 8 1.0 - 1.0 +0/=2/-0
50.00%
Lc0v18 11089 - Stockfish_18090410_x64_bmi2 0.5 - 1.5 +0/=1/-1
25.00%

_ The gaunlet with FEN:_
Leela had 11.0/14

Lc0v18 11089     - Andscacs 0.94                     1.5 - 0.5    +1/=1/-0  

75.00%
Lc0v18 11089 - Booot 6.3.1 1.5 - 0.5 +1/=1/-0
75.00%
Lc0v18 11089 - Ethereal 11.00-x64-pext 1.5 - 0.5 +1/=1/-0
75.00%
Lc0v18 11089 - Fire 7.1 1.5 - 0.5 +1/=1/-0
75.00%
Lc0v18 11089 - Fizbo2x64_bmi2 2.0 - 0.0 +2/=0/-0
100.00%
Lc0v18 11089 - Stockfish 8 1.5 - 0.5 +1/=1/-0
75.00%
Lc0v18 11089 - Stockfish_18090410_x64_bmi2 1.5 - 0.5 +1/=1/-0
75.00%

So in this particular variant, Leela doesn’t seem to be affected much by the FEN issue she has. She even got a point more with FEN!