Komodo 2-pawn odds match: round 1

[Vinvin]

After succeeding in a lot of pieces exchanges, Neuman gives a pawn to enter in completely winning endgame !
Score 1-2 !

[Event "Computer chess game"]
[Site "jesse"]
[Date "2015.08.04"]
[Round "3"]
[White "Komodo"]
[Black "GM Neuman"]
[Result "0-1"]
[BlackElo "3300"]
[Time "15:40:31"]
[WhiteElo "2464"]
[TimeControl "2700+15"]
[SetUp "1"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/8/2N5/PP1PP1PP/R1BQKBNR w KQkq - 0 1"]
[Termination "unterminated"]

1. Nf3 {-1.09/24 84} (1. Nf3 Nf6 2. d4 c5 3. e3 Nc6 4. Bd2 e6 5. Bd3 a6 6.
O-O b5 7. Qe1 c4 8. Bc2 Bd6 9. e4 Qb6 10. Be3 dxe4 11. Nxe4 Nxe4 12. Bxe4
f5 13. Bxc6+ Qxc6 14. Ne5 Qd5 15. Qg3 O-O 16. Rf3 Bb7) 1. .. Nf6 2. e3
{-1.21/26 189} (2. e3 c5 3. d4 Nc6 4. Bb5 e6 5. O-O Bd7 6. Bd2 a6 7. Bxc6
Bxc6 8. Be1 Bd6 9. dxc5 Bxc5 10. Nd4 O-O 11. Bh4 Rc8 12. Bxf6 gxf6 13. Qh5
Bxd4 14. exd4 Qb6 15. Qg4+ Kh8 16. b3 f5 17. Qf4 Rg8 18. Qe5+) 2. .. g6 3.
Be2 {-1.21/25 143} (3. Be2 Bg7 4. O-O O-O 5. Kh1 Nc6 6. d4 a6 7. Bd2 Qd6 8.
Be1 Ng4 9. Bd2 Bf5 10. h3 Nf6 11. Be1 Ne4 12. Nxe4 dxe4 13. Nd2 e5 14. g4
Be6 15. Nxe4 Bd5 16. Bf3 Bxe4 17. Bxe4 exd4 18. exd4) 3. .. Bg7 4. b4
{-1.25/26 160} (4. b4 a6 5. Qa4+ c6 6. O-O b5 7. Qb3 a5 8. Ba3 a4 9. Qc2
Bf5 10. Qc1 O-O 11. Nd1 Qd6 12. Bb2 Nbd7 13. Nf2 e5 14. a3 Ne4 15. g4 Be6
16. Nxe4 dxe4 17. Ng5 Bd5 18. Qc2 Nf6 19. Qc3 Rfe8) 4. .. O-O 5. Bb2
{-1.36/27 80} (5. Bb2 Nc6 6. b5 Na5 7. Qa4 c5 8. bxc6 Nxc6 9. Nb5 Bg4 10.
Rc1 Qd7 11. Qb3 Bxf3 12. Bxf3 Rac8 13. O-O a6 14. Na3 e5 15. d3 Rfe8 16.
Nc2 e4 17. dxe4 dxe4 18. Be2 Na5 19. Qa3 Qd2 20. Nd4 Nd5 21. Rxc8) 5. ..
Bg4 6. O-O {-1.16/24 47} (6. O-O c6 7. h3 Bxf3 8. Bxf3 Nbd7 9. b5 Ne5 10.
bxc6 bxc6 11. Na4 Nxf3+ 12. Rxf3 Qd6 13. Rc1 Rab8 14. Rc2 Nd7 15. Bxg7 Kxg7
16. d4 f5 17. Qc1 Qb4 18. Nc5 Nxc5 19. Rxc5 Qb1 20. Rf1 Qxc1) 6. .. Nbd7 7.
Na4 {-1.21/25 80} (7. Na4 c6 8. Rc1 Qc7 9. h3 Bf5 10. Nc5 Qg3 11. Qe1 Ne4
12. Bxg7 Kxg7 13. Nxb7 Bxh3 14. Qxg3 Nxg3 15. Rfe1 Bg4 16. Na5 Nxe2+ 17.
Rxe2 Ne5 18. Nxc6 Bxf3 19. gxf3 Nxc6 20. Rxc6 Rfb8 21. e4 d4 22. Rc7 Rxb4
23. Rxe7 Rb1+) 7. .. c6 8. Rc1 {-1.21/25 38} (8. Rc1 Qc7) 8. .. Bxf3 9.
gxf3 {-1.24/26 50} (9. gxf3 Rc8 10. Kh1 e5 11. Nc5 Nxc5 12. bxc5 d4 13. Bc4
Nd5 14. Qb3 Rb8 15. exd4 exd4 16. f4 Re8 17. f5 Qe7 18. Rf3 Qh4 19. Rg1 Re1
20. Bxd5 cxd5 21. fxg6 hxg6 22. Qxd5 Rf8 23. Rg3 Re2 24. R3g2 Qe4) 9. .. e5
10. Kh1 {-1.49/30 127} (10. Kh1 Rc8 11. Nc5 Nxc5 12. bxc5 d4 13. Qb3 Nd5
14. Rc4 b5 15. cxb6 Qxb6 16. Ba3 Rfd8 17. Qc2 Qa6 18. Bc5 Nb6 19. Bxb6 axb6
20. a4 Qb7 21. e4 Qd7 22. Bd3 Qe6 23. Qb3 c5 24. a5 bxa5 25. Qb5 Rb8 26.
Qxa5 Bf8 27. Rcc1 Qb6 28. Qxb6 Rxb6 29. Kg2) 10. .. a6 11. Nc5 {-1.37/29
98} (11. Nc5 Qe7 12. Nxb7 Qxb4 13. Qb3 a5 14. Bc3 Qxb3 15. axb3 Rfe8 16.
Bd3 c5 17. Ra1 Reb8 18. Rxa5 Rxa5 19. Nxa5 e4 20. Bc2 exf3 21. Rxf3 Rb6 22.
Kg2 Ne4 23. Bxe4 dxe4 24. Rf1 Bxc3 25. dxc3 Ne5 26. Kg3 f5 27. Rd1 Nd3 28.
Ra1 Kg7 29. Ra3 Ra6 30. Nc4 Rxa3 31. Nxa3) 11. .. Nxc5 12. bxc5 {-1.38/25
27} (12. bxc5 Qe7 13. Qb3 Rad8 14. Rg1 Rfe8 15. Bc3 Rd7 16. Ba5 e4 17. f4
d4 18. Rg5 Nd5 19. a3 f5 20. Rg3 Kh8 21. Rh3 Qf6 22. Bc4 dxe3 23. dxe3 Qb2
24. Rg1 Qxb3 25. Bxb3 Bf8) 12. .. Qe7 13. Qb3 {-1.46/30 65} (13. Qb3 Rad8
14. Bc3 Rfe8 15. Rg1 Rd7 16. Ba5 e4 17. f4 d4 18. Rg3 Nd5 19. Bg4 f5 20.
Be2 Kh8 21. Rgg1 Qf6 22. Bc4 Qe6 23. Rg3 Qe7 24. Bf1 Rg8 25. Rg1 Qe6 26.
Rg3 Bf6 27. Rg2 Re8 28. Be2 Qe7 29. Bf1) 13. .. Rad8 14. Bc3 {-1.67/32 251}
(14. Bc3 Rd7) 14. .. Rd7 15. Ba5 {-1.92/29 9} (15. Ba5 e4 16. f4 d4 17. Rg1
Rd5 18. Qb4 Re8 19. Bc4 Rh5 20. Bf1 d3 21. Rg3 Qe6 22. Qb3 Qd7 23. Kg1 b5
24. Bg2 Nd5 25. Qd1 Kh8 26. Qb3 f5 27. Bf1 Qb7 28. Bg2 Qf7 29. Kh1 Qe7 30.
Bh3 Qa7 31. Kg1) 15. .. Kh8 16. Rb1 {-1.88/27 128} (16. Rb1 e4 17. Bb4 exf3
18. Bxf3 Ne4 19. d4 f5 20. Bg2 Nf6 21. Bd2 Bh6 22. Rf3 Ne4 23. Be1 g5 24.
Rf1 g4 25. Qd3 Qe8 26. Rb2 Bg5 27. Re2 Rdf7 28. Rg1 Qe7 29. Ba5 Qe6 30. Rb2
Re8 31. Re1 f4 32. exf4) 16. .. Ng8 17. e4 {-1.63/26 84} (17. e4 Bh6 18.
exd5 Bf4 19. Rg1 cxd5 20. Rg2 Rc8 21. Bb6 Qe6 22. Rbg1 Qh3 23. Bd1 Ne7 24.
Qb2 Nf5 25. Qc3 Re7 26. Ba4 Nd4 27. Rf2 f6 28. Rgg2 Qh4 29. Bb3 Nxb3 30.
axb3 e4 31. fxe4) 17. .. Nf6 18. Qe3 {-1.71/26 39} (18. Qe3 Re8 19. d3 Nh5
20. Rb2 d4 21. Qc1 Nf4 22. Rg1 Bh6 23. Qc4 Qe6 24. Qxe6 Nxe6 25. Rc2 Be3
26. Rb1 Kg7 27. Bf1 Rde7 28. Bh3 f5 29. exf5 Nf4 30. Bf1 gxf5 31. Rcb2 Rb8
32. Rb6 e4 33. fxe4) 18. .. Nh5 19. exd5 {-1.69/29 43} (19. exd5 Rxd5 20.
Bc4 Rd7 21. d3 Qh4 22. Rg1 Re8 23. Rg4 Qe7 24. Rg2 f5 25. Rgg1 Qh4 26. Be1
Qf4 27. Qe2 e4 28. fxe4 Qe5 29. Qe3 fxe4 30. Qxe4 Qxe4+ 31. dxe4 Nf6 32. e5
Rxe5 33. Bg3 Re8 34. Rge1 Ne4 35. Bh4 Bh6 36. Bf1 Bd2 37. Re2 g5 38. Be1
Bxe1 39. Rbxe1) 19. .. Rxd5 20. Bc4 {-1.76/30 53} (20. Bc4 Rd7 21. d3 Qh4
22. Qf2 Qxf2 23. Rxf2 Rb8 24. Bd2 Nf4 25. Rb3 Bf8 26. Be3 Nd5 27. Bxd5 Rxd5
28. Rc2 Kg7 29. Bf2 Be7 30. Kg2 Bg5 31. a4 Bf4 32. Rcb2 Rd7 33. Bg3 h5 34.
Bf2 Rbd8 35. Rxb7 Rxb7 36. Rxb7 Rxd3 37. Rb6 Rd2) 20. .. Rd7 21. d3
{-1.87/28 23} (21. d3 Re8 22. Rb3 Qh4 23. Rfb1 Ng3+ 24. Kg1 Nf5 25. Qf2
Qxf2+ 26. Kxf2 Ree7 27. R3b2 Bh6 28. Re2 Bf4 29. h3 Nd4 30. Reb2 Kg7 31.
Re1 Nb5 32. Bb4 f6 33. Reb1 h5 34. a4 Nd4 35. Bc3 Nf5 36. Re2 Re8 37. Rg1
Red8 38. Re4 Nd4 39. Ba5) 21. .. Nf4 22. Rb3 {-1.90/30 108} (22. Rb3 Qg5
23. Qg1 Qxg1+ 24. Kxg1 Rb8 25. Re1 f6 26. Bd2 Bf8 27. Be3 Rbd8 28. Kh1 Nxd3
29. Bxd3 Rxd3 30. Rxb7 Rc3 31. Rf7 Bxc5 32. Bxc5 Rxc5 33. Re2 Rd6 34. Kg2
a5 35. Rb2 Rb5 36. Rc2 Kg8 37. Rc7 Rbd5 38. Kg3 Rd2 39. R7xc6 Kf7 40. Rc7+
Ke6 41. Rxd2 Rxd2 42. a4 Rd4 43. Rxh7 Rxa4) 22. .. Nd5 23. Qg1 {-1.86/29
79} (23. Qg1 Bh6 24. d4 e4 25. Bxd5 Rxd5 26. fxe4 Rg5 27. Qf2 Qxe4+ 28. Qf3
Qc2 29. Rf2 Qc1+ 30. Rf1 Qc4 31. Qd3 Qxd3 32. Rxd3 Rd5 33. Rb3 b5 34. Rf6
Rc8 35. Rd6 Rf5 36. Rb2 Kg7 37. Re2 Bg5 38. Kg2 Bf4 39. Rd7 Rh5 40. Ree7
Rf8 41. h3 Rg5+ 42. Kf3 Rf5 43. Kg2 Bb8 44. Re3 Rg5+ 45. Kf3 Rg3+ 46. Kf2)
23. .. Bh6 24. Re1 (Re1 f5 Qf2 Re8 Rb2 Bg5 Qg1 Bh4 Reb1 Nf4 Rd2 Bg5 Rdd1
Nd5 Re1 Bh6 Qf2 Bf8 Qb2 Bg7 Qf2 Bf6 Qg1 Nf4 Rb3 Bh4 Reb1 Ne2 Qe3 Nd4 R3b2
Bg5 Qf2 Qxc5 Bb6 Qe7 a4 Bh4 Qe3 f4 Qe4) {-2.16/31 138} Bf4 25. Re2 (Re2 Qh4
Be1 Qh5 d4 exd4 Qxd4 f6 Bxd5 Be5 Qg4 Qxg4 fxg4 cxd5 Kg2 Rc8 Bb4 Kg7 Rd3 d4
h3 Rh8 Be1 Kf7 Bg3 Re8 Re1 Rd5 Rb1 Rxc5 Bxe5 Rc2 Kg3 fxe5) {-2.07/28 53} f6
26. Bc3 (Bc3 Re8 Ba1 Qf8 Re4 f5 Re2 Qh6 Bxd5 cxd5 Bd4 Rde7 Bb2 Kg8 Rb4 Rc8
Qf2 Rcc7 Rb6 Qh3 Re1 d4 Ba3 Rc6 Rb2 g5 Qg2 Qxg2 Kxg2 g4 fxg4 fxg4 h3 gxh3
Kxh3 Rh6 Kg2) {-1.84/28 57} Nxc3 27. Rxc3 (Rxc3 f5 Rb3 Qh4 Rb4 Qh3 Qg2 Qh6
Qg1 g5 Rb1 Qh3 Rf1 Rg7 Rg2 Rd8 Qf2 h5 a4 g4 fxg4 hxg4 Qb2 Rd4 a5 Qh6 Qb6
Rh7 Qb2 Re7 Re1 Rg7 Be6) {-2.14/27 23} Rc8 28. Rb3 (Rb3 Re8 Rb1 Bh6 Qf2 Bf8
Reb2 f5 Qe2 Bh6 a4 e4 dxe4 fxe4 Qxe4 Qxe4 fxe4 Rxe4 Bf1 Ree7 Bh3 Rc7 Kg2
Be3 Rc2 Kg7 a5 h5 Rd1 Rf7 Rd3 Rce7 Re2 Bxc5 Rxe7 Bxe7) {-2.01/30 36} Rcc7
29. Reb2 (Reb2 Bh6 Rc3 Rd4 Qb1 f5 Rcb3 Rdd7 Qg1 Qg5 Re2 Qxg1 Kxg1 Re7 Kf2
Rcd7 Re1 Kg7 h4 Kf6 Re2 Bf8 Ra3 Rd4 Kg3 Bh6 Rc3 Kg7 Kh3 Bf4 Bb3 Rdd7 Rg2
Kf6 Bc4 Rg7) {-2.49/29 62} Kg7 30. a4 (a4 Kh8 a5 Bh6 Qf2 Bf8 Qe1 Qxc5 Re2
Rd4 Reb2 Qe7 Qb1 Rdd7 Qe1 Qg7 Be6 Rd4 Qb1 Rd6 Bc4 Rdd7 Be6 Re7 Bc4 Qh6 Qf1
Qf4 Qf2 Qc1 Kg2 Bh6 Rc2 Qe3 Qf1 Qc5 Qe1 f5) {-2.42/32 78} f5 31. Rb1 (Rb1
Qg5 Qxg5 Bxg5 Re1 Bf4 Re2 Kf6 Kg2 h5 h3 g5 Kf2 g4 fxg4 hxg4 hxg4 fxg4 Rb6
Rd4 Rb1 Rh7 Rg1 Kg5 Kg2 Rh2 Kf1 Rh3 Kf2 Rf3 Kg2 Rg3 Kf1 Rxg1 Kxg1 Rd7 a5
Rh7 Be6 Rh3 Rg2) {-2.61/30 116} Qg5 32. Qxg5 (Qxg5 Bxg5 Re1 Bf4 Kg2 Kf6 Rb2
h5 h3 Rd4 Kf2 g5 Rbe2 Re7 a5 Rh7 Rh1 Rhd7 Rb1 g4 hxg4 hxg4 fxg4 e4 Rh1 Rg7
Rh5 Rxg4 Rh7 Bg3 Ke3 exd3 Bxd3 Bh4 Kd2 Rg3 Kc2 Rdxd3 Rxh4 Ra3 Rb4 Rg7 Re8
Rxa5 Rf8 Ke6 Re8 Kd5 Kc3) {-2.92/34 49} Bxg5 33. Re1 (Re1 Be7 Rxe5 Bf6 Re8
Re7 Rxe7 Rxe7 Kg2 Bd4 a5 Bxc5 h4 Kf6 Kg3 Be3 Ra3 Bd2 Kh3 Be1 Ra1 Bc3 Ra3
Be5 Kg2 Bb2 Rb3 Bd4 Kg3 Be5 Kg2 g5 h5 Bd4 Rb4 Bc3 Ra4 Kg7 Kf2 Bd4)
{-3.09/34 32} Kf6 34. Rb2 (Rb2 Bf4 Kg2 g5 h3 Rd4 Rbe2 h5 Kf2 Re7 a5 Rh7 Rh1
Rhd7 Rb1 g4 hxg4 hxg4 fxg4 e4 Rh1 Rg7 Rh5 Rxg4 Rh7 Bg3 Ke3 exd3 Bxd3 Bh4
Kd2 Rg3 Kc2 Rdxd3 Rxh4 Ra3 Kd2 Rxa5 Rh6 Kg5 Rh7 Ra2 Ke1 Rxe2 Kxe2 Rb3)
{-2.96/32 27} Bf4 35. h3 (h3 g5 Kg2 h5 Kf2 Rh7 Rh1 Rcg7 Re2 g4 hxg4 hxg4
Rxh7 g3 Kf1 Rxh7 Kg2 Rh2 Kf1 Rh5 Kg2 Bc1 Kxg3 Be3 Kg2 Bxc5 Ra2 Rg5 Kh2 Bd4
a5 Rg7 Re2 f4 Ra2 Rg5 Ra4 Rg3 Ra3 Bg1 Kh1 Be3 Rb3) {-2.95/33 35} a5 36. Kg2
(Kg2 Rd4 Ree2 Bc1 Rb3 Rcd7 Kg3 Bf4 Kf2 Bh6 Reb2 e4 fxe4 fxe4 Rxb7 exd3 Rxd7
Rxd7 Rb6 d2 Bb3 d1=R Rxc6 Ke5 Bxd1 Rxd1 Rc7 Rd2 Kf3 Rd3 Kg2 Bf4 Rxh7 Rc3
Re7 Kf5 Rf7 Ke4 Re7 Be5 Re6 Rc2 Kf1 Rxc5 Ke2 g5 Ra6 Kf4 Ra8 Kg3 Rg8 Bc3)
{-3.32/32 39} g5 37. Rbe2 (Rbe2 Rd4 Rc2 Rcd7 Rb1 Rg7 Re2 h5 Rg1 Rh7 Rh1 Rc7
Rb1 Rcd7 Rc2 Rh7 Rh1 Be3 Re2 f4 Kf1 Rdd7 Rxe3 fxe3 Ke2 Rd4 Kxe3 Rhd7 Kf2
Rf4 Ke3) {-3.21/30 27} h5 38. Rb1 (Rb1 Rd4 Kf2 Rcd7 Rc2 Rh7 Rh1 Rg7 Rg1 Re7
Re2 Red7 Rb1 Rc7 Kg2 Rg7 Rh1 Rh7 Rc2 Rhd7 Rb1 Rg7 Rh1 Re7 Rb1 Rdd7 Re2 Rh7
Rh1 Rhg7 Rd1 Rge7) {-3.37/31 32} g4 39. hxg4 (hxg4 hxg4 Rh1 Rh7 fxg4 Rxh1
g5 Bxg5 Kxh1 Rd7 Kg2 Bh6 Kf3 Bf8 Rh2 Rd4 Ke2 Bxc5 Rh6 Kg5 Rh7 e4 Rg7 Kf6
Rxb7 f4 Rf7 Kg5 Ba6 f3 Ke1 e3 Rxf3 Rxa4 d4 Bxd4 Rf8 Ra1 Ke2 Ra2 Kd3 Rd2 Ke4
c5 Rg8 Kf6 Kf3 a4 Bc4 a3 Rf8 Ke7 Ra8) {-3.91/33 66} hxg4 40. Rh1 (Rh1 Rh7
Ree1 Bd2 Rd1 gxf3 Kxf3 Bb4 Bg8 Rxh1 Rxh1 Bxc5 Ke2 Rd7 Bc4 Rd4 Rh3 Rg4 Rh6
Kg5 Rh7 Rg2 Kf3 Rf2 Kg3 f4 Kh3 Rb2 Rg7 Kf5 Rf7 Kg6 Kg4 Rg2 Kf3 Rg3 Ke2 Re3
Kf1 Rh3 Ke2 Bd4 Rf8 Rh7 Kf3 Kg5 Kg2 Bc5 Rg8 Kf6 Bb3 Rd7 Bc4 Bd4) {-4.17/37
88} Rh7 41. Ree1 (Ree1 Bd2 Rd1 gxf3 Kxf3 Bb4 Bg8 Rxh1 Rxh1 Bxc5 Bc4 Rd7 Rh6
Kg5 Re6 Bd4 Re8 f4 Rf8 Rh7 Kg2 Bc5 Rf7 Rxf7 Bxf7 Kf6 Bb3 Bd4 Kf3 Ke7 Bd1
Kd6 Ke2 Be3 Ke1 Kc5 Bg4 Bd4 Kd2 Kb4 Bd1 Be3 Ke1 Kc3 Bg4 Bd4 Ke2 Kb4)
{-4.29/37 33} g3 42. Bb3 (Bb3 Rcd7 Bc4 Rdg7 Bb3 Bd2 Rd1 Rxh1 Rxh1 Be3 Rh8
Bxc5 d4 exd4 f4 Bd6 Rh4 d3 Rh6 Ke7 Re6 Kd7 Re3 Bxf4 Rxd3 Bd6 Rd2 Rh7 Ra2
Rh2 Kf1 Rh4 Kg2 f4 Kf3 Rh1 Bc4 Kc7 Be2 Rg1 Bd3 Rc1 Kg2 Re1 Be2 Rb1 Bd3 Rc1
Be2 Re1) {-5.12/35 59} Rcd7 43. Bc4 (Bc4 Rdg7 Bb3 Rxh1 Rxh1 Be3 Rh8 Bxc5 d4
exd4 f4 Bd6 Rh4 d3 Rh6 Ke7 Re6 Kd7 Re3 Bxf4 Rxd3 Bd6 Bc2 f4 Rd2 Re7 Bf5 Kc7
Ra2 Re3 Bg4 Rd3 Bf3 Rb3 Be2 Rb1 Bf3 Bc5 Kh3 Rb4 Kg4 Kd6 Rd2 Ke6 Ra2 Be3 Bg2
Kd6 Kf3 c5 Ra1) {-5.79/37 49} Rh2+ 44. Rxh2 (Rxh2 gxh2 Rd1 Rd4 Bb3 Kg5 Bc2
Rd5 Kh1 Rxc5 Bb3 b5 Bf7 Kf6 Bb3 Rc3 Ba2 bxa4 Bc4 a3 Ra1 Rc2 Rd1 Rb2 Ra1 Kg5
Be6 Rd2 Bc4 Rf2 Re1 Rxf3 Kg2 Kg4 Kh1 Bg3 Rc1 Bf4) {-7.20/30 14} gxh2 45.
Kh1 (Kh1 Kg5 Re2 Kh4 Rg2 Rd4 Rg8 e4 Rh8 Kg5 Rg8 Kh5 fxe4 fxe4 Bf7 Kh4 Rh8
Kg5 dxe4 Rd1 Kg2 Rg1 Kf3 h1=Q Rxh1 Rxh1 Be6 Be5 Bc8 Rb1 Be6 Ra1 Bc8 Ra3 Ke2
Rb3 Kd2 Kf4 Kc2 Rb2 Kd3 Rb4 Kc2 Kxe4 Kd1 Rb2 Kc1 Kd5 Bf5 Kxc5 Bc2 Rxc2 Kxc2
Kb4 Kd2) {-8.32/34 12} Rg7 46. Bb3 (Bb3 Rg5 Bc2 Kg6 Bd1 Kh5 Be2 Kh4 Rb1 Rg7
Ra1 Kg3 Bf1 Kxf3 Bh3 e4 Rf1 Kg3 Bg2 Re7 Re1 e3 Re2 Kg4 d4 Rd7 d5 cxd5 Bf1
d4 Rg2 Kf3 c6 bxc6 Bd3 Rd5 Rc2 Re5 Re2 Rc5 Re1 Rc3 Rf1 Kg3 Rd1 c5 Be2 Kf2
Rf1 Kxe2 Rxf4 d3 Kxh2 d2 Rxf5 d1=Q Rf6 Kd2 Rh6 e2 Re6 e1=Q Rxe1) {-16.33/40
56} Ke7 47. Bc4 (Bc4 Rh7 Re2 Rh3 Rb2 Rxf3 Rxb7 Kf6 Rb1 e4 dxe4 fxe4 Bg8 Rc3
Bb3 Rxc5 Rf1 Kg5 Rb1 Rc3 Kg2 Rd3 Bc4 Rg3 Kh1 Rc3 Ba6 Rc5 Rd1 Kf6 Kg2 Rc2
Kh1 Ke5 Bc8 Bg3 Bg4 e3 Rd8 Kf4 Bh5 e2 Rf8 Ke4 Bf3 Kd3 Bxe2 Rxe2 Rf3 Re3
Rf1) {-7.80/35 108} Kf6 48. Bb3 (Bb3) {0.00/80 42} Kg5 49. Be6 (Be6 Re7 Bc8
Rc7 Be6 Kf6 Ba2 Re7 Re2 Rd7 Bb1 Rd4 Bc2 Rb4 Re1 Rb2 Bb1 Rb3 Bc2 Rc3 Bb1 Kg5
Kg2 Rxc5 Rd1 Rc3 Kh1 b5 Rf1 b4 Rd1 b3 Rf1 b2 Kg2 Rc1 Rh1 Kf6 Ba2 Ra1 Bb1
Rxa4 Rd1 Rd4 Rh1 a4 Kf2 a3 Ke2) {-8.82/32 7} Re7 50. Bc8 (Bc8 Rc7 Be6 Kf6
Ba2 Re7 Bc4 Rd7 Re2 Rd4 Re1 Kg5 Rb1 Rd7 Rb2 Kh4 Rg2 Rd4 Rg8 e4 Rh8 Kg3 fxe4
fxe4 Rxh2 Kf3 Rg2 Rd8 Kg1 Be3 Kf1 Rh8 Rg3 Kxg3 Ke2 Bxc5 dxe4 Kf4 Kd3 Rh3
Kd2 Ra3 Be6 Rxa4 Bb3 Rd4 Kc2 Rxe4 Kd2 Rb4 Bc2 Ke5 Bh7 a4 Bd3 Rf4) {-9.71/37
1} Rc7 51. Be6 (Be6 Kf6 Ba2 Re7 Bc4 Rd7 Re2 Rd4 Re1 Kg5 Rb1 Rd7 Be6 Rxd3
Bc8 Rxf3 Bxb7 Rc3 Bxc6 Rxc5 Bb5 e4 Rf1 Rc2 Ba6 Rb2 Bc4 e3 Rf3 Rb4 Bd3 Rxa4
Rf1 Ra2 Re1 Rd2 Bf1 a4 Be2 Bg3 Rf1 a3 Kg2 Rxe2 Kxg3 a2 Rc1 Rf2 Kh3 f4 Ra1
e2 Rh1 Rf1) {-11.29/36 21} Bg3 52. Re2 (Re2 Rg7 Ra2 Kf4 Bc8 Kxf3 Bxf5 Bf4
Ra1 Ke3 Rc1 Rg5 Be4 Rg4 Rb1 Rg7 Rf1 Kd4 Ra1 Kxc5 Bg2 Kd4 Ra2 Kxd3 Ra3 Kc4
Ra1 Kc5 Ra2 Kd4 Ra1 Rg3 Rd1 Rd3 Ra1 Kc5 Be4 Rc3 Bg2) {-6.51/31 10} Kf4 53.
Rc2 (Rc2 Re7 Bc8 Rg7 Rc3 Kxf3 Bxf5 Bf4 Be4 Ke2 Bg2 Kd2 Ra3 Kc2 Ra2 Kxd3 Ra3
Kd4 Ra2 Kxc5 Rb2 Kc4 Rb6 Kc3 Rb1 Kd4 Rf1 Kd3 Rd1 Kc4 Ra1 Kb4 Rd1 Kc5 Ra1
Rg3 Ra2 Rc3 Ra1 b5 axb5) {-10.40/34 31} Re7 54. Bc8 (Bc8 Rg7 Kxf3 Bxf5 Bf4
Kf2 Ke3 Kd4 Be4 Kxc5 Bg2 b5 d4 exd4 Rf7 Be4 bxa4 Bg6 Rf6 Bd3 a3 Kb4 Kg2 c5
Be4 c4 Kc5 Rh6 Kh1 Rh3 a2 Re3 Kg2 Rxe4 d3 axa5 Kd4 a1=Q c3 Ke3 c2 Be5)
{-10.40/34 4} Rc7 55. Be6 (Be6) {0.00/80 34} Kxf3 56. Bxf5 (Bxf5 Rf7 Bg6
Rf4 Be4 Ke3 Kg2 Rg4 Rb2 Bf4 Kh3 Rg7 Kh4 Kd4 Rc2 h1=R Bxh1 Rh7 Kg4 Rxh1 Rb2
Rh7 Rb3 Rd7 Rb1 Kxd3 Kf5 Kc4 Ke6 Rg7 Kf5 Rh7 Kg6 Rd7 Kf6 Kxc5 Rb2 Kd4 Rb1
Ke4 Rb2 Be3 Ke6) {-7.30/30 8} Ke3 57. Ra2 (Ra2 Bf4 Rg2 Kd4 Be4 Kxc5 Rc2 Kd4
Kg2 Ke3 Rc3 Rg7 Kh3 Rh7 Kg4 h1=Q Bxh1 Rxh1 Rb3 Rh7 Rc3 Kd4 Rb3 Rd7 Kf5 Kc5
d4 Kxd4 Rb1 Kc5 Ke6 Rh7 Rd1 b5 axb5 cxb5 Rd8 Kb4 Ra8 a4 Kd5 a3 Ke6 Kb3)
{-8.51/30 15} Rf7 58. Rg2 (Rg2 Bf4 Be4 Rd7 Bf5 Rc7 Be4 Kd4 Rg8 Kxc5 Kg2 b5
Ra8 Kb4 Kh3 Kxa4 Kg4 c5 Kf5 Kb4 Ke6 a4 Kd6 Rh7 Kc6 a3 Kd5 h1=Q Bxh1 Rxh1
Ra7 Rh6 Rg7 Ra6 Rg1 a2 Ra1 Ra3 Ke6 Kc3 Kd5 b4 Kxc5 b3 Kb5 Kxd3) {-10.05/31
7} Rxf5 59. Rxg3+ (Rxg3 Kd4 Kxh2 Kxc5 Rg2 b5 Rc2 Kb4 axb5 cxb5 Rb2 Kc5 Rc2
Kd4 Ra2 a4 Ra3 Kc5 Ra1 Kb4 Rb1 Ka5 Kg3 a3 Re1 Ka4 Kg4 Rf8 Kh5 a2 Ra1 Kb3
Kh6 b4 Kg7 Rf4 Kg6 Ka3 Kg5 b3 Re1 Ka4) {-7.23/26 11} Kd4 60. Kxh2 (Kxh2
Kxc5 Rg2 b5 Rc2 Kb4 axb5 cxb5 Rb2 Kc5 Rc2 Kd4 Ra2 a4 Kg3 Kxd3 Ra1 e4 Kg4
Rc5 Rd1 Kc2 Rg1 e3 Kf3 Kd2 Rg2 Kd3 Rg7 Rf5 Kg4 e2 Kxf5 e1=Q Rd7 Kc3 Rc7 Kb2
Rg7 a3 Rg2 Kc3 Kg4 Qe6 Kf3) {-8.99/25 4} Kxc5 61. Rg2 (Rg2 b5 Rb2 b4 Kg2
Kd5 Rb1 Kd4 Rb3 Rf4 Kg3 Rf1 Kg4 Rd1 Kf5 Rxd3 Rb2 e4 Rc2 Kd5 Kf4 b3 Re2 c5
Rb2 c4 Kf5 Kd4 Ke6 c3 Rxb3 c2 Rb7 c1=Q Rd7 Ke3 Re7 Ke2 Rf7 Rf3 Rh7)
{-12.05/28 28} Rf3 62. Rd2 (Rd2 b5 Kg2 Rf5 Rc2 Kb4 Rc1 Kxa4 Ra1 Kb4 Kh3 a4
Kg4 Rf4 Kg5 a3 d4 Rxd4 Kf6 e4 Rb1 Kc3 Rc1 Kb2 Re1 a2 Re2 Kb3 Re3 Ka4 Re1 c5
Rc1 e3 Re1 e2 Kf7 Kb3 Kg8 c4 Rxe2 a1=Q Re6 c3 Rg6 c2 Rg3 Kb2 Rg6)
{-15.00/29 33} Rf4 63. Rc2+ (Rc2 Kd4 Rb2 Kxd3 Rxb7 Rxa4 Kg3 Rc4 Ra7 a4 Kf3
Rf4 Kg2 Kc2 Kg3 Rd4 Kg2 Kb3 Kg3 a3 Rb7 Rb4 Rf7 a2 Rf3 Kc4 Kg4 a1=Q Kg5 Qg1
Kf6 Qd4 Rh3 e4 Ke6 e3 Rg3 e2 Kf5 Rb5 Ke6 e1=Q Re3 Qdxe3 Kd7 Qf3)
{-250.00/35 15} Kd6 64. Rb2 (Rb2 Rb4 Ra2 Kc5 Rc2 Kd5 Kg3 Rxa4 Kf2 Rf4 Ke1
b5 Ra2 a4 Ke2 Rh4 Ra1 Rh2 Kf3 Rh3 Kg4 Rxd3 Kf5 a3 Rb1 e4 Kf4 a2 Ra1 Rf3 Kg4
Ra3 Rd1 Kc4 Rc1 Kb3 Ra1 Kb2 Rf1 a1=Q Rf2 Kb3 Kf5 e3 Rf3 Qg1 Ke5 Kb2 Kd6 e2
Rxa3) {-15.95/31 17} Rb4 65. Ra2 (Ra2 Kc5 Re2 Kd5 Kg3 Rxa4 Kf2 b5 Ke1 Rf4
Kd2 b4 Kc2 a4 Rg2 Rf1 Kb2 Rf3 Kc2 Kd4 Rg4 Rf4 Rg2 c5 Rh2 a3 Rh8 Rf2 Kb3 Rb2
Ka4 a2 Rh1 Rb1 Rh6 a1=Q Kb5 Rb2 Rd6 Ke3 Rb6 Rc2 d4 cxd4 Rg6 d3 Rg4 b3)
{-17.30/32 0} Rb3 66. Rd2 (Rd2 Kd5 Kg2 Rb4 Kf3 Rxa4 Rh2 b5 Rh7 b4 Rb7 Kd4
Rg7 Kxd3 Rg4 Ra2 Rg6 c5 Rd6 Kc4 Ke4 b3 Rd1 b2 Rb1 a4 Kxe5 a3 Re1 Ra1 Re4
Kb5 Kd6 b1=Q Re7 Kc4 Kc7 Qd3 Re8 Qh7 Kc8 Rd1 Rd8 Rxd8 Kxd8) {-14.25/27 21}
Ra3 67. Rb2 (Rb2 Kc7 Rf2 Rxa4 Rf7 Kb6 Kg3 Rd4 Rh7 a4 Kf3 Rxd3 Ke4 Rd5 Rh2
Kc5 Rh7 b5 Ra7 Kd6 Kf5 Rd4 Ra8 Rf4 Kg6 e4 Rd8 Kc5 Rd1 a3 Rc1 Kd5 Rd1 Ke5
Rd2 b4 Rc2 b3 Rxc6 a2 Ra6 Rf1 Ra3 Rg1 Kf7 a1=Q) {-9.43/28 12} Kc7 68. Rg2
(Rg2 Rxa4 Rg7 Kb6 Kg2 Rd4 Kf3 Rxd3 Ke4 Rd5 Rg1 a4 Rb1 Kc7 Rh1 b5 Ra1 Kd6
Kf5 Rd4 Kg6 e4 Kf5 Kd5 Kf6 Rd3 Ke7 a3 Kd8 b4 Kc7 e3 Kb6 e2 Re1 a2 Rxe2 b3
Rh2 a1=Q Rh5 Qe5 Rh1 b2 Ka6 Ra3 Kb7 Qe7 Kb8) {-14.45/30 16} Rxd3 69. Rg7+
(Rg7 Rd7 Rg5 Kd6 Kg3 Rf7 Kg4 Rf4 Kh5 Rxa4 Kg6 Rd4 Kf7 a4 Rg6 Kd5 Rg2 b5 Ra2
e4 Ke7 e3 Kd7 Re4 Re2 a3 Kc7 b4 Rc2 e2 Kb6 e1=Q Rc5 Kd4 Rxc6 b3 Rd6 Kc3 Kc7
a2 Rd7 b2 Kd8 Re8 Kc7 Qe5 Rd6) {-19.02/28 9} Kb6 70. Kg2 (Kg2 Rd4 Kf3 Rxa4
Rg5 e4 Ke3 c5 Kf4 Rd4 Rg2 a4 Ke5 Kc6 Ra2 b5 Ra3 Rd3 Ra2 e3 Rg2 c4 Rg6 Kc5
Rg2 a3 Ke4 b4 Rg5 Kc6 Rg6 Kb5 Rg5 Ka4 Re5 a2 Re8 e2 Kf5 b3 Rxe2 a1=Q Re7 b2
Rc7 Qf1 Ke6 Re3 Kd7 Qf5 Kd8) {-16.72/30 16} Rd4 71. Kf3 (Rxa4 Re7 Rd4 Rxe5
Rd5 Re1 Kc5 Rc1 Kd6 Ra1 b5 Rxa5 b4 Ra1 b3 Kc5 Rc1 Kb4 Rb1 b2 c5 c4 Rf1 b1=Q
Rf4 Rd3 Rxc4) {-16.72/30 27} Rxa4 72. Rg5 (Rg5 Rd4 Rxe5 a4 Ke3 Rd5 Re4 Ka5
Kf2 b5 Kg3 Kb6 Re8 Kc5 Kh4 b4 Rb8 a3 Re8 b3 Rb8 a2 Ra8 b2 Rxa2 b1=Q Ra5 Kd6
Rxd5) {-250.00/44 2} e4+ 73. Ke3 (Ke3 c5 Rg6 Kb5 Rg7 b6 Kf4 Rd4 Ke5 a4 Rg2
Kc4 Rc2 Kb3 Rc1 a3 Kf6 a2 Re1 b5 Ke7 b4 Ke6 Kc4 Ke7 b3 Ra1 b2 Rxa2 b1=Q Rh2
e3 Rh5 Qg6 Rxc5 Kxc5) {-250.00/37 10} Rc4 74. Rg6 (Rg6 a4 Rg8 c5 Ra8 Rd4
Kf4 Kb5 Rg8 Kc4 Kf5 b5 Ra8 Kb3 Ra7 a3 Rg7 a2 Rg1 e3 Kg6 e2 Re1 Rd1 Rxe2
a1=Q Re3 Qc3 Rxc3 Kxc3) {-250.00/43 13} a4 75. Rg8 (Rg8 Kb5 Ra8 Kb4 Rb8 a3
Rxb7 Ka4 Ra7 Kb3 Ra8 a2 Rb8 Rb4 Ra8 Ra4 Rb8 Kc4 Rc8 c5 Kf4 a1=Q Kf5 Ra6
Rxc5 Kxc5) {-250.00/37 2} Kb5 76. Rg5+ (Rg5 Kb4 Rg1 a3 Ra1 Kb3 Rb1 Kc2 Rxb7
a2 Ra7 Kb3 Kf4 Rc1 Rb7 Kc4 Rb2 a1=Q Re2 Qd4 Rxe4 Qxe4) {-250.00/44 6} Kb4
77. Rg7 (Rg7 a3 Rxb7 Kc3 Ra7 Kb3 Ra6 a2 Kf2 e3 Ke2 Ra4 Rb6 Ka3 Rxc6 a1=Q
Rd6 Qd4 Rd5 Qxd5) {-250.00/42 4} b5 78. Rg1 (Rg1 a3 Kf4 a2 Re1 Kb3 Rh1 Kb2
Rh2 Rc2 Rxc2 Kxc2 Ke3 c5 Kxe4) {-250.00/42 9} a3 79. Kf4 (Kf4 a2 Re1 Kb3
Kf5 Ra4 Re3 Kc2 Re1 a1=Q Rxa1 Rxa1 Kxe4) {-250.00/46 3} a2 80. Ke5 (Ke5 Kb3
Kd6 Kb2 Rg2 Rc2 Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/47 3} Kb3 81. Kd6 (Kd6
Kb2 Rg2 Rc2 Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/49 2} Kb2 82. Rg2+ (Rg2 Rc2
Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/53 4} Rc2 83. Rxc2+ (Rxc2 Kxc2 Kd7 a1=Q
Kxc6) {-250.00/70 47} Kxc2 84. Kd7 (Kd7 a1=Q Kxc6) {-250.00/70 0} a1=Q 85.
Kxc6 {-250.00/70 0} Qe5 0-1

[lkaufman]

Two draw, one loose. Interesting games. What do you think about this result Larry ?

Well, I was hoping for an even score, which doesn't look likely now. But this is really a huge handicap. After 172 self-play games of new Komodo giving itself this two pawn handicap at 15' + 5" on two threads, the score is zero wins, two draws, and 170 losses! That's a 901 elo gap. The elo value of a handicap rises with the time control, so at 45' + 15" it would probably exceed one thousand elo points. But of course there are various factors that might make this number either too low or too high for a human match.

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

[Guenther]

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

Well, playing against itself there shouldn't be much surprises in the search tree(s), which could overcome a substantial handicap in the first move, no?
(BTW how did you randomize the openings in selfplay over 170 games?
Is there enough variety, just because of 2 threads, or did you observe a lot of quasi-doubles until a real winning point in the game?)

Guenther

P.S. I find those handicap games vs. Humans very interesting. Thanks for this!

[JJJ]

With these result, can you estimate better the FIDE elo of Komodo ?

[lkaufman]

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

Well, playing against itself there shouldn't be much surprises in the search tree(s), which could overcome a substantial handicap in the first move, no?
(BTW how did you randomize the openings in selfplay over 170 games?
Is there enough variety, just because of 2 threads, or did you observe a lot of quasi-doubles until a real winning point in the game?)

Guenther

P.S. I find those handicap games vs. Humans very interesting. Thanks for this!

I used six short sensible opening sequences, and relied on the unpredictability of MP for more variety. I think this works well enough; games usually diverge quite early. I agree about more surprises in the search tree when playing unrelated engines; I'm just surprised at the huge effect this has.

[Laskos]

Two draw, one loose. Interesting games. What do you think about this result Larry ?

Well, I was hoping for an even score, which doesn't look likely now. But this is really a huge handicap. After 172 self-play games of new Komodo giving itself this two pawn handicap at 15' + 5" on two threads, the score is zero wins, two draws, and 170 losses! That's a 901 elo gap. The elo value of a handicap rises with the time control, so at 45' + 15" it would probably exceed one thousand elo points. But of course there are various factors that might make this number either too low or too high for a human match.

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

On such large a span, the rating model becomes important. FIDE uses Gaussian instead of logistic used in all engine ratings. Gaussian ELO difference for 0.994 is 700 instead of 900 for logistic. Then, the error margins (2SD) for your result after 170 games are about 160 ELO points lower bound and even larger higher bound, similar to a result after 10 games of an even match. I first matched time control odds to get fairly equal strength, then let them play a balanced match. No large ELO differences, but I had to assume that I know the value ELO (logistic or Gaussian, doesn't matter for small intervals) of doubling time control from ultra-fast to long TC, connected doubling by doubling, but I did derive that 2 years ago or so. I agree that the handicap can be larger than 600-650 ELO points I derived, but I doubt it is larger than 700 ELO points of connected small intervals, i.e. maybe Gaussian points or maybe logistic. Once I found for several engines that they obey rather a Gaussian than a logistic, but it's not established yet.

[carldaman]

After succeeding in a lot of pieces exchanges, Neuman gives a pawn to enter in completely winning endgame !
Score 1-2 !

[pgn]

[Event "Computer chess game"]
[Site "jesse"]
[Date "2015.08.04"]
[Round "3"]
[White "Komodo"]
[Black "GM Neuman"]
[Result "0-1"]
[BlackElo "3300"]
[Time "15:40:31"]
[WhiteElo "2464"]
[TimeControl "2700+15"]
[SetUp "1"]
[FEN "rnbqkbnr/ppp1pppp/8/3p4/8/2N5/PP1PP1PP/R1BQKBNR w KQkq - 0 1"]
[Termination "unterminated"]

1. Nf3 {-1.09/24 84} (1. Nf3 Nf6 2. d4 c5 3. e3 Nc6 4. Bd2 e6 5. Bd3 a6 6.
O-O b5 7. Qe1 c4 8. Bc2 Bd6 9. e4 Qb6 10. Be3 dxe4 11. Nxe4 Nxe4 12. Bxe4
f5 13. Bxc6+ Qxc6 14. Ne5 Qd5 15. Qg3 O-O 16. Rf3 Bb7) 1. .. Nf6 2. e3
{-1.21/26 189} (2. e3 c5 3. d4 Nc6 4. Bb5 e6 5. O-O Bd7 6. Bd2 a6 7. Bxc6
Bxc6 8. Be1 Bd6 9. dxc5 Bxc5 10. Nd4 O-O 11. Bh4 Rc8 12. Bxf6 gxf6 13. Qh5
Bxd4 14. exd4 Qb6 15. Qg4+ Kh8 16. b3 f5 17. Qf4 Rg8 18. Qe5+) 2. .. g6 3.
Be2 {-1.21/25 143} (3. Be2 Bg7 4. O-O O-O 5. Kh1 Nc6 6. d4 a6 7. Bd2 Qd6 8.
Be1 Ng4 9. Bd2 Bf5 10. h3 Nf6 11. Be1 Ne4 12. Nxe4 dxe4 13. Nd2 e5 14. g4
Be6 15. Nxe4 Bd5 16. Bf3 Bxe4 17. Bxe4 exd4 18. exd4) 3. .. Bg7 4. b4
{-1.25/26 160} (4. b4 a6 5. Qa4+ c6 6. O-O b5 7. Qb3 a5 8. Ba3 a4 9. Qc2
Bf5 10. Qc1 O-O 11. Nd1 Qd6 12. Bb2 Nbd7 13. Nf2 e5 14. a3 Ne4 15. g4 Be6
16. Nxe4 dxe4 17. Ng5 Bd5 18. Qc2 Nf6 19. Qc3 Rfe8) 4. .. O-O 5. Bb2
{-1.36/27 80} (5. Bb2 Nc6 6. b5 Na5 7. Qa4 c5 8. bxc6 Nxc6 9. Nb5 Bg4 10.
Rc1 Qd7 11. Qb3 Bxf3 12. Bxf3 Rac8 13. O-O a6 14. Na3 e5 15. d3 Rfe8 16.
Nc2 e4 17. dxe4 dxe4 18. Be2 Na5 19. Qa3 Qd2 20. Nd4 Nd5 21. Rxc8) 5. ..
Bg4 6. O-O {-1.16/24 47} (6. O-O c6 7. h3 Bxf3 8. Bxf3 Nbd7 9. b5 Ne5 10.
bxc6 bxc6 11. Na4 Nxf3+ 12. Rxf3 Qd6 13. Rc1 Rab8 14. Rc2 Nd7 15. Bxg7 Kxg7
16. d4 f5 17. Qc1 Qb4 18. Nc5 Nxc5 19. Rxc5 Qb1 20. Rf1 Qxc1) 6. .. Nbd7 7.
Na4 {-1.21/25 80} (7. Na4 c6 8. Rc1 Qc7 9. h3 Bf5 10. Nc5 Qg3 11. Qe1 Ne4
12. Bxg7 Kxg7 13. Nxb7 Bxh3 14. Qxg3 Nxg3 15. Rfe1 Bg4 16. Na5 Nxe2+ 17.
Rxe2 Ne5 18. Nxc6 Bxf3 19. gxf3 Nxc6 20. Rxc6 Rfb8 21. e4 d4 22. Rc7 Rxb4
23. Rxe7 Rb1+) 7. .. c6 8. Rc1 {-1.21/25 38} (8. Rc1 Qc7) 8. .. Bxf3 9.
gxf3 {-1.24/26 50} (9. gxf3 Rc8 10. Kh1 e5 11. Nc5 Nxc5 12. bxc5 d4 13. Bc4
Nd5 14. Qb3 Rb8 15. exd4 exd4 16. f4 Re8 17. f5 Qe7 18. Rf3 Qh4 19. Rg1 Re1
20. Bxd5 cxd5 21. fxg6 hxg6 22. Qxd5 Rf8 23. Rg3 Re2 24. R3g2 Qe4) 9. .. e5
10. Kh1 {-1.49/30 127} (10. Kh1 Rc8 11. Nc5 Nxc5 12. bxc5 d4 13. Qb3 Nd5
14. Rc4 b5 15. cxb6 Qxb6 16. Ba3 Rfd8 17. Qc2 Qa6 18. Bc5 Nb6 19. Bxb6 axb6
20. a4 Qb7 21. e4 Qd7 22. Bd3 Qe6 23. Qb3 c5 24. a5 bxa5 25. Qb5 Rb8 26.
Qxa5 Bf8 27. Rcc1 Qb6 28. Qxb6 Rxb6 29. Kg2) 10. .. a6 11. Nc5 {-1.37/29
98} (11. Nc5 Qe7 12. Nxb7 Qxb4 13. Qb3 a5 14. Bc3 Qxb3 15. axb3 Rfe8 16.
Bd3 c5 17. Ra1 Reb8 18. Rxa5 Rxa5 19. Nxa5 e4 20. Bc2 exf3 21. Rxf3 Rb6 22.
Kg2 Ne4 23. Bxe4 dxe4 24. Rf1 Bxc3 25. dxc3 Ne5 26. Kg3 f5 27. Rd1 Nd3 28.
Ra1 Kg7 29. Ra3 Ra6 30. Nc4 Rxa3 31. Nxa3) 11. .. Nxc5 12. bxc5 {-1.38/25
27} (12. bxc5 Qe7 13. Qb3 Rad8 14. Rg1 Rfe8 15. Bc3 Rd7 16. Ba5 e4 17. f4
d4 18. Rg5 Nd5 19. a3 f5 20. Rg3 Kh8 21. Rh3 Qf6 22. Bc4 dxe3 23. dxe3 Qb2
24. Rg1 Qxb3 25. Bxb3 Bf8) 12. .. Qe7 13. Qb3 {-1.46/30 65} (13. Qb3 Rad8
14. Bc3 Rfe8 15. Rg1 Rd7 16. Ba5 e4 17. f4 d4 18. Rg3 Nd5 19. Bg4 f5 20.
Be2 Kh8 21. Rgg1 Qf6 22. Bc4 Qe6 23. Rg3 Qe7 24. Bf1 Rg8 25. Rg1 Qe6 26.
Rg3 Bf6 27. Rg2 Re8 28. Be2 Qe7 29. Bf1) 13. .. Rad8 14. Bc3 {-1.67/32 251}
(14. Bc3 Rd7) 14. .. Rd7 15. Ba5 {-1.92/29 9} (15. Ba5 e4 16. f4 d4 17. Rg1
Rd5 18. Qb4 Re8 19. Bc4 Rh5 20. Bf1 d3 21. Rg3 Qe6 22. Qb3 Qd7 23. Kg1 b5
24. Bg2 Nd5 25. Qd1 Kh8 26. Qb3 f5 27. Bf1 Qb7 28. Bg2 Qf7 29. Kh1 Qe7 30.
Bh3 Qa7 31. Kg1) 15. .. Kh8 16. Rb1 {-1.88/27 128} (16. Rb1 e4 17. Bb4 exf3
18. Bxf3 Ne4 19. d4 f5 20. Bg2 Nf6 21. Bd2 Bh6 22. Rf3 Ne4 23. Be1 g5 24.
Rf1 g4 25. Qd3 Qe8 26. Rb2 Bg5 27. Re2 Rdf7 28. Rg1 Qe7 29. Ba5 Qe6 30. Rb2
Re8 31. Re1 f4 32. exf4) 16. .. Ng8 17. e4 {-1.63/26 84} (17. e4 Bh6 18.
exd5 Bf4 19. Rg1 cxd5 20. Rg2 Rc8 21. Bb6 Qe6 22. Rbg1 Qh3 23. Bd1 Ne7 24.
Qb2 Nf5 25. Qc3 Re7 26. Ba4 Nd4 27. Rf2 f6 28. Rgg2 Qh4 29. Bb3 Nxb3 30.
axb3 e4 31. fxe4) 17. .. Nf6 18. Qe3 {-1.71/26 39} (18. Qe3 Re8 19. d3 Nh5
20. Rb2 d4 21. Qc1 Nf4 22. Rg1 Bh6 23. Qc4 Qe6 24. Qxe6 Nxe6 25. Rc2 Be3
26. Rb1 Kg7 27. Bf1 Rde7 28. Bh3 f5 29. exf5 Nf4 30. Bf1 gxf5 31. Rcb2 Rb8
32. Rb6 e4 33. fxe4) 18. .. Nh5 19. exd5 {-1.69/29 43} (19. exd5 Rxd5 20.
Bc4 Rd7 21. d3 Qh4 22. Rg1 Re8 23. Rg4 Qe7 24. Rg2 f5 25. Rgg1 Qh4 26. Be1
Qf4 27. Qe2 e4 28. fxe4 Qe5 29. Qe3 fxe4 30. Qxe4 Qxe4+ 31. dxe4 Nf6 32. e5
Rxe5 33. Bg3 Re8 34. Rge1 Ne4 35. Bh4 Bh6 36. Bf1 Bd2 37. Re2 g5 38. Be1
Bxe1 39. Rbxe1) 19. .. Rxd5 20. Bc4 {-1.76/30 53} (20. Bc4 Rd7 21. d3 Qh4
22. Qf2 Qxf2 23. Rxf2 Rb8 24. Bd2 Nf4 25. Rb3 Bf8 26. Be3 Nd5 27. Bxd5 Rxd5
28. Rc2 Kg7 29. Bf2 Be7 30. Kg2 Bg5 31. a4 Bf4 32. Rcb2 Rd7 33. Bg3 h5 34.
Bf2 Rbd8 35. Rxb7 Rxb7 36. Rxb7 Rxd3 37. Rb6 Rd2) 20. .. Rd7 21. d3
{-1.87/28 23} (21. d3 Re8 22. Rb3 Qh4 23. Rfb1 Ng3+ 24. Kg1 Nf5 25. Qf2
Qxf2+ 26. Kxf2 Ree7 27. R3b2 Bh6 28. Re2 Bf4 29. h3 Nd4 30. Reb2 Kg7 31.
Re1 Nb5 32. Bb4 f6 33. Reb1 h5 34. a4 Nd4 35. Bc3 Nf5 36. Re2 Re8 37. Rg1
Red8 38. Re4 Nd4 39. Ba5) 21. .. Nf4 22. Rb3 {-1.90/30 108} (22. Rb3 Qg5
23. Qg1 Qxg1+ 24. Kxg1 Rb8 25. Re1 f6 26. Bd2 Bf8 27. Be3 Rbd8 28. Kh1 Nxd3
29. Bxd3 Rxd3 30. Rxb7 Rc3 31. Rf7 Bxc5 32. Bxc5 Rxc5 33. Re2 Rd6 34. Kg2
a5 35. Rb2 Rb5 36. Rc2 Kg8 37. Rc7 Rbd5 38. Kg3 Rd2 39. R7xc6 Kf7 40. Rc7+
Ke6 41. Rxd2 Rxd2 42. a4 Rd4 43. Rxh7 Rxa4) 22. .. Nd5 23. Qg1 {-1.86/29
79} (23. Qg1 Bh6 24. d4 e4 25. Bxd5 Rxd5 26. fxe4 Rg5 27. Qf2 Qxe4+ 28. Qf3
Qc2 29. Rf2 Qc1+ 30. Rf1 Qc4 31. Qd3 Qxd3 32. Rxd3 Rd5 33. Rb3 b5 34. Rf6
Rc8 35. Rd6 Rf5 36. Rb2 Kg7 37. Re2 Bg5 38. Kg2 Bf4 39. Rd7 Rh5 40. Ree7
Rf8 41. h3 Rg5+ 42. Kf3 Rf5 43. Kg2 Bb8 44. Re3 Rg5+ 45. Kf3 Rg3+ 46. Kf2)
23. .. Bh6 24. Re1 (Re1 f5 Qf2 Re8 Rb2 Bg5 Qg1 Bh4 Reb1 Nf4 Rd2 Bg5 Rdd1
Nd5 Re1 Bh6 Qf2 Bf8 Qb2 Bg7 Qf2 Bf6 Qg1 Nf4 Rb3 Bh4 Reb1 Ne2 Qe3 Nd4 R3b2
Bg5 Qf2 Qxc5 Bb6 Qe7 a4 Bh4 Qe3 f4 Qe4) {-2.16/31 138} Bf4 25. Re2 (Re2 Qh4
Be1 Qh5 d4 exd4 Qxd4 f6 Bxd5 Be5 Qg4 Qxg4 fxg4 cxd5 Kg2 Rc8 Bb4 Kg7 Rd3 d4
h3 Rh8 Be1 Kf7 Bg3 Re8 Re1 Rd5 Rb1 Rxc5 Bxe5 Rc2 Kg3 fxe5) {-2.07/28 53} f6
26. Bc3 (Bc3 Re8 Ba1 Qf8 Re4 f5 Re2 Qh6 Bxd5 cxd5 Bd4 Rde7 Bb2 Kg8 Rb4 Rc8
Qf2 Rcc7 Rb6 Qh3 Re1 d4 Ba3 Rc6 Rb2 g5 Qg2 Qxg2 Kxg2 g4 fxg4 fxg4 h3 gxh3
Kxh3 Rh6 Kg2) {-1.84/28 57} Nxc3 27. Rxc3 (Rxc3 f5 Rb3 Qh4 Rb4 Qh3 Qg2 Qh6
Qg1 g5 Rb1 Qh3 Rf1 Rg7 Rg2 Rd8 Qf2 h5 a4 g4 fxg4 hxg4 Qb2 Rd4 a5 Qh6 Qb6
Rh7 Qb2 Re7 Re1 Rg7 Be6) {-2.14/27 23} Rc8 28. Rb3 (Rb3 Re8 Rb1 Bh6 Qf2 Bf8
Reb2 f5 Qe2 Bh6 a4 e4 dxe4 fxe4 Qxe4 Qxe4 fxe4 Rxe4 Bf1 Ree7 Bh3 Rc7 Kg2
Be3 Rc2 Kg7 a5 h5 Rd1 Rf7 Rd3 Rce7 Re2 Bxc5 Rxe7 Bxe7) {-2.01/30 36} Rcc7
29. Reb2 (Reb2 Bh6 Rc3 Rd4 Qb1 f5 Rcb3 Rdd7 Qg1 Qg5 Re2 Qxg1 Kxg1 Re7 Kf2
Rcd7 Re1 Kg7 h4 Kf6 Re2 Bf8 Ra3 Rd4 Kg3 Bh6 Rc3 Kg7 Kh3 Bf4 Bb3 Rdd7 Rg2
Kf6 Bc4 Rg7) {-2.49/29 62} Kg7 30. a4 (a4 Kh8 a5 Bh6 Qf2 Bf8 Qe1 Qxc5 Re2
Rd4 Reb2 Qe7 Qb1 Rdd7 Qe1 Qg7 Be6 Rd4 Qb1 Rd6 Bc4 Rdd7 Be6 Re7 Bc4 Qh6 Qf1
Qf4 Qf2 Qc1 Kg2 Bh6 Rc2 Qe3 Qf1 Qc5 Qe1 f5) {-2.42/32 78} f5 31. Rb1 (Rb1
Qg5 Qxg5 Bxg5 Re1 Bf4 Re2 Kf6 Kg2 h5 h3 g5 Kf2 g4 fxg4 hxg4 hxg4 fxg4 Rb6
Rd4 Rb1 Rh7 Rg1 Kg5 Kg2 Rh2 Kf1 Rh3 Kf2 Rf3 Kg2 Rg3 Kf1 Rxg1 Kxg1 Rd7 a5
Rh7 Be6 Rh3 Rg2) {-2.61/30 116} Qg5 32. Qxg5 (Qxg5 Bxg5 Re1 Bf4 Kg2 Kf6 Rb2
h5 h3 Rd4 Kf2 g5 Rbe2 Re7 a5 Rh7 Rh1 Rhd7 Rb1 g4 hxg4 hxg4 fxg4 e4 Rh1 Rg7
Rh5 Rxg4 Rh7 Bg3 Ke3 exd3 Bxd3 Bh4 Kd2 Rg3 Kc2 Rdxd3 Rxh4 Ra3 Rb4 Rg7 Re8
Rxa5 Rf8 Ke6 Re8 Kd5 Kc3) {-2.92/34 49} Bxg5 33. Re1 (Re1 Be7 Rxe5 Bf6 Re8
Re7 Rxe7 Rxe7 Kg2 Bd4 a5 Bxc5 h4 Kf6 Kg3 Be3 Ra3 Bd2 Kh3 Be1 Ra1 Bc3 Ra3
Be5 Kg2 Bb2 Rb3 Bd4 Kg3 Be5 Kg2 g5 h5 Bd4 Rb4 Bc3 Ra4 Kg7 Kf2 Bd4)
{-3.09/34 32} Kf6 34. Rb2 (Rb2 Bf4 Kg2 g5 h3 Rd4 Rbe2 h5 Kf2 Re7 a5 Rh7 Rh1
Rhd7 Rb1 g4 hxg4 hxg4 fxg4 e4 Rh1 Rg7 Rh5 Rxg4 Rh7 Bg3 Ke3 exd3 Bxd3 Bh4
Kd2 Rg3 Kc2 Rdxd3 Rxh4 Ra3 Kd2 Rxa5 Rh6 Kg5 Rh7 Ra2 Ke1 Rxe2 Kxe2 Rb3)
{-2.96/32 27} Bf4 35. h3 (h3 g5 Kg2 h5 Kf2 Rh7 Rh1 Rcg7 Re2 g4 hxg4 hxg4
Rxh7 g3 Kf1 Rxh7 Kg2 Rh2 Kf1 Rh5 Kg2 Bc1 Kxg3 Be3 Kg2 Bxc5 Ra2 Rg5 Kh2 Bd4
a5 Rg7 Re2 f4 Ra2 Rg5 Ra4 Rg3 Ra3 Bg1 Kh1 Be3 Rb3) {-2.95/33 35} a5 36. Kg2
(Kg2 Rd4 Ree2 Bc1 Rb3 Rcd7 Kg3 Bf4 Kf2 Bh6 Reb2 e4 fxe4 fxe4 Rxb7 exd3 Rxd7
Rxd7 Rb6 d2 Bb3 d1=R Rxc6 Ke5 Bxd1 Rxd1 Rc7 Rd2 Kf3 Rd3 Kg2 Bf4 Rxh7 Rc3
Re7 Kf5 Rf7 Ke4 Re7 Be5 Re6 Rc2 Kf1 Rxc5 Ke2 g5 Ra6 Kf4 Ra8 Kg3 Rg8 Bc3)
{-3.32/32 39} g5 37. Rbe2 (Rbe2 Rd4 Rc2 Rcd7 Rb1 Rg7 Re2 h5 Rg1 Rh7 Rh1 Rc7
Rb1 Rcd7 Rc2 Rh7 Rh1 Be3 Re2 f4 Kf1 Rdd7 Rxe3 fxe3 Ke2 Rd4 Kxe3 Rhd7 Kf2
Rf4 Ke3) {-3.21/30 27} h5 38. Rb1 (Rb1 Rd4 Kf2 Rcd7 Rc2 Rh7 Rh1 Rg7 Rg1 Re7
Re2 Red7 Rb1 Rc7 Kg2 Rg7 Rh1 Rh7 Rc2 Rhd7 Rb1 Rg7 Rh1 Re7 Rb1 Rdd7 Re2 Rh7
Rh1 Rhg7 Rd1 Rge7) {-3.37/31 32} g4 39. hxg4 (hxg4 hxg4 Rh1 Rh7 fxg4 Rxh1
g5 Bxg5 Kxh1 Rd7 Kg2 Bh6 Kf3 Bf8 Rh2 Rd4 Ke2 Bxc5 Rh6 Kg5 Rh7 e4 Rg7 Kf6
Rxb7 f4 Rf7 Kg5 Ba6 f3 Ke1 e3 Rxf3 Rxa4 d4 Bxd4 Rf8 Ra1 Ke2 Ra2 Kd3 Rd2 Ke4
c5 Rg8 Kf6 Kf3 a4 Bc4 a3 Rf8 Ke7 Ra8) {-3.91/33 66} hxg4 40. Rh1 (Rh1 Rh7
Ree1 Bd2 Rd1 gxf3 Kxf3 Bb4 Bg8 Rxh1 Rxh1 Bxc5 Ke2 Rd7 Bc4 Rd4 Rh3 Rg4 Rh6
Kg5 Rh7 Rg2 Kf3 Rf2 Kg3 f4 Kh3 Rb2 Rg7 Kf5 Rf7 Kg6 Kg4 Rg2 Kf3 Rg3 Ke2 Re3
Kf1 Rh3 Ke2 Bd4 Rf8 Rh7 Kf3 Kg5 Kg2 Bc5 Rg8 Kf6 Bb3 Rd7 Bc4 Bd4) {-4.17/37
88} Rh7 41. Ree1 (Ree1 Bd2 Rd1 gxf3 Kxf3 Bb4 Bg8 Rxh1 Rxh1 Bxc5 Bc4 Rd7 Rh6
Kg5 Re6 Bd4 Re8 f4 Rf8 Rh7 Kg2 Bc5 Rf7 Rxf7 Bxf7 Kf6 Bb3 Bd4 Kf3 Ke7 Bd1
Kd6 Ke2 Be3 Ke1 Kc5 Bg4 Bd4 Kd2 Kb4 Bd1 Be3 Ke1 Kc3 Bg4 Bd4 Ke2 Kb4)
{-4.29/37 33} g3 42. Bb3 (Bb3 Rcd7 Bc4 Rdg7 Bb3 Bd2 Rd1 Rxh1 Rxh1 Be3 Rh8
Bxc5 d4 exd4 f4 Bd6 Rh4 d3 Rh6 Ke7 Re6 Kd7 Re3 Bxf4 Rxd3 Bd6 Rd2 Rh7 Ra2
Rh2 Kf1 Rh4 Kg2 f4 Kf3 Rh1 Bc4 Kc7 Be2 Rg1 Bd3 Rc1 Kg2 Re1 Be2 Rb1 Bd3 Rc1
Be2 Re1) {-5.12/35 59} Rcd7 43. Bc4 (Bc4 Rdg7 Bb3 Rxh1 Rxh1 Be3 Rh8 Bxc5 d4
exd4 f4 Bd6 Rh4 d3 Rh6 Ke7 Re6 Kd7 Re3 Bxf4 Rxd3 Bd6 Bc2 f4 Rd2 Re7 Bf5 Kc7
Ra2 Re3 Bg4 Rd3 Bf3 Rb3 Be2 Rb1 Bf3 Bc5 Kh3 Rb4 Kg4 Kd6 Rd2 Ke6 Ra2 Be3 Bg2
Kd6 Kf3 c5 Ra1) {-5.79/37 49} Rh2+ 44. Rxh2 (Rxh2 gxh2 Rd1 Rd4 Bb3 Kg5 Bc2
Rd5 Kh1 Rxc5 Bb3 b5 Bf7 Kf6 Bb3 Rc3 Ba2 bxa4 Bc4 a3 Ra1 Rc2 Rd1 Rb2 Ra1 Kg5
Be6 Rd2 Bc4 Rf2 Re1 Rxf3 Kg2 Kg4 Kh1 Bg3 Rc1 Bf4) {-7.20/30 14} gxh2 45.
Kh1 (Kh1 Kg5 Re2 Kh4 Rg2 Rd4 Rg8 e4 Rh8 Kg5 Rg8 Kh5 fxe4 fxe4 Bf7 Kh4 Rh8
Kg5 dxe4 Rd1 Kg2 Rg1 Kf3 h1=Q Rxh1 Rxh1 Be6 Be5 Bc8 Rb1 Be6 Ra1 Bc8 Ra3 Ke2
Rb3 Kd2 Kf4 Kc2 Rb2 Kd3 Rb4 Kc2 Kxe4 Kd1 Rb2 Kc1 Kd5 Bf5 Kxc5 Bc2 Rxc2 Kxc2
Kb4 Kd2) {-8.32/34 12} Rg7 46. Bb3 (Bb3 Rg5 Bc2 Kg6 Bd1 Kh5 Be2 Kh4 Rb1 Rg7
Ra1 Kg3 Bf1 Kxf3 Bh3 e4 Rf1 Kg3 Bg2 Re7 Re1 e3 Re2 Kg4 d4 Rd7 d5 cxd5 Bf1
d4 Rg2 Kf3 c6 bxc6 Bd3 Rd5 Rc2 Re5 Re2 Rc5 Re1 Rc3 Rf1 Kg3 Rd1 c5 Be2 Kf2
Rf1 Kxe2 Rxf4 d3 Kxh2 d2 Rxf5 d1=Q Rf6 Kd2 Rh6 e2 Re6 e1=Q Rxe1) {-16.33/40
56} Ke7 47. Bc4 (Bc4 Rh7 Re2 Rh3 Rb2 Rxf3 Rxb7 Kf6 Rb1 e4 dxe4 fxe4 Bg8 Rc3
Bb3 Rxc5 Rf1 Kg5 Rb1 Rc3 Kg2 Rd3 Bc4 Rg3 Kh1 Rc3 Ba6 Rc5 Rd1 Kf6 Kg2 Rc2
Kh1 Ke5 Bc8 Bg3 Bg4 e3 Rd8 Kf4 Bh5 e2 Rf8 Ke4 Bf3 Kd3 Bxe2 Rxe2 Rf3 Re3
Rf1) {-7.80/35 108} Kf6 48. Bb3 (Bb3) {0.00/80 42} Kg5 49. Be6 (Be6 Re7 Bc8
Rc7 Be6 Kf6 Ba2 Re7 Re2 Rd7 Bb1 Rd4 Bc2 Rb4 Re1 Rb2 Bb1 Rb3 Bc2 Rc3 Bb1 Kg5
Kg2 Rxc5 Rd1 Rc3 Kh1 b5 Rf1 b4 Rd1 b3 Rf1 b2 Kg2 Rc1 Rh1 Kf6 Ba2 Ra1 Bb1
Rxa4 Rd1 Rd4 Rh1 a4 Kf2 a3 Ke2) {-8.82/32 7} Re7 50. Bc8 (Bc8 Rc7 Be6 Kf6
Ba2 Re7 Bc4 Rd7 Re2 Rd4 Re1 Kg5 Rb1 Rd7 Rb2 Kh4 Rg2 Rd4 Rg8 e4 Rh8 Kg3 fxe4
fxe4 Rxh2 Kf3 Rg2 Rd8 Kg1 Be3 Kf1 Rh8 Rg3 Kxg3 Ke2 Bxc5 dxe4 Kf4 Kd3 Rh3
Kd2 Ra3 Be6 Rxa4 Bb3 Rd4 Kc2 Rxe4 Kd2 Rb4 Bc2 Ke5 Bh7 a4 Bd3 Rf4) {-9.71/37
1} Rc7 51. Be6 (Be6 Kf6 Ba2 Re7 Bc4 Rd7 Re2 Rd4 Re1 Kg5 Rb1 Rd7 Be6 Rxd3
Bc8 Rxf3 Bxb7 Rc3 Bxc6 Rxc5 Bb5 e4 Rf1 Rc2 Ba6 Rb2 Bc4 e3 Rf3 Rb4 Bd3 Rxa4
Rf1 Ra2 Re1 Rd2 Bf1 a4 Be2 Bg3 Rf1 a3 Kg2 Rxe2 Kxg3 a2 Rc1 Rf2 Kh3 f4 Ra1
e2 Rh1 Rf1) {-11.29/36 21} Bg3 52. Re2 (Re2 Rg7 Ra2 Kf4 Bc8 Kxf3 Bxf5 Bf4
Ra1 Ke3 Rc1 Rg5 Be4 Rg4 Rb1 Rg7 Rf1 Kd4 Ra1 Kxc5 Bg2 Kd4 Ra2 Kxd3 Ra3 Kc4
Ra1 Kc5 Ra2 Kd4 Ra1 Rg3 Rd1 Rd3 Ra1 Kc5 Be4 Rc3 Bg2) {-6.51/31 10} Kf4 53.
Rc2 (Rc2 Re7 Bc8 Rg7 Rc3 Kxf3 Bxf5 Bf4 Be4 Ke2 Bg2 Kd2 Ra3 Kc2 Ra2 Kxd3 Ra3
Kd4 Ra2 Kxc5 Rb2 Kc4 Rb6 Kc3 Rb1 Kd4 Rf1 Kd3 Rd1 Kc4 Ra1 Kb4 Rd1 Kc5 Ra1
Rg3 Ra2 Rc3 Ra1 b5 axb5) {-10.40/34 31} Re7 54. Bc8 (Bc8 Rg7 Kxf3 Bxf5 Bf4
Kf2 Ke3 Kd4 Be4 Kxc5 Bg2 b5 d4 exd4 Rf7 Be4 bxa4 Bg6 Rf6 Bd3 a3 Kb4 Kg2 c5
Be4 c4 Kc5 Rh6 Kh1 Rh3 a2 Re3 Kg2 Rxe4 d3 axa5 Kd4 a1=Q c3 Ke3 c2 Be5)
{-10.40/34 4} Rc7 55. Be6 (Be6) {0.00/80 34} Kxf3 56. Bxf5 (Bxf5 Rf7 Bg6
Rf4 Be4 Ke3 Kg2 Rg4 Rb2 Bf4 Kh3 Rg7 Kh4 Kd4 Rc2 h1=R Bxh1 Rh7 Kg4 Rxh1 Rb2
Rh7 Rb3 Rd7 Rb1 Kxd3 Kf5 Kc4 Ke6 Rg7 Kf5 Rh7 Kg6 Rd7 Kf6 Kxc5 Rb2 Kd4 Rb1
Ke4 Rb2 Be3 Ke6) {-7.30/30 8} Ke3 57. Ra2 (Ra2 Bf4 Rg2 Kd4 Be4 Kxc5 Rc2 Kd4
Kg2 Ke3 Rc3 Rg7 Kh3 Rh7 Kg4 h1=Q Bxh1 Rxh1 Rb3 Rh7 Rc3 Kd4 Rb3 Rd7 Kf5 Kc5
d4 Kxd4 Rb1 Kc5 Ke6 Rh7 Rd1 b5 axb5 cxb5 Rd8 Kb4 Ra8 a4 Kd5 a3 Ke6 Kb3)
{-8.51/30 15} Rf7 58. Rg2 (Rg2 Bf4 Be4 Rd7 Bf5 Rc7 Be4 Kd4 Rg8 Kxc5 Kg2 b5
Ra8 Kb4 Kh3 Kxa4 Kg4 c5 Kf5 Kb4 Ke6 a4 Kd6 Rh7 Kc6 a3 Kd5 h1=Q Bxh1 Rxh1
Ra7 Rh6 Rg7 Ra6 Rg1 a2 Ra1 Ra3 Ke6 Kc3 Kd5 b4 Kxc5 b3 Kb5 Kxd3) {-10.05/31
7} Rxf5 59. Rxg3+ (Rxg3 Kd4 Kxh2 Kxc5 Rg2 b5 Rc2 Kb4 axb5 cxb5 Rb2 Kc5 Rc2
Kd4 Ra2 a4 Ra3 Kc5 Ra1 Kb4 Rb1 Ka5 Kg3 a3 Re1 Ka4 Kg4 Rf8 Kh5 a2 Ra1 Kb3
Kh6 b4 Kg7 Rf4 Kg6 Ka3 Kg5 b3 Re1 Ka4) {-7.23/26 11} Kd4 60. Kxh2 (Kxh2
Kxc5 Rg2 b5 Rc2 Kb4 axb5 cxb5 Rb2 Kc5 Rc2 Kd4 Ra2 a4 Kg3 Kxd3 Ra1 e4 Kg4
Rc5 Rd1 Kc2 Rg1 e3 Kf3 Kd2 Rg2 Kd3 Rg7 Rf5 Kg4 e2 Kxf5 e1=Q Rd7 Kc3 Rc7 Kb2
Rg7 a3 Rg2 Kc3 Kg4 Qe6 Kf3) {-8.99/25 4} Kxc5 61. Rg2 (Rg2 b5 Rb2 b4 Kg2
Kd5 Rb1 Kd4 Rb3 Rf4 Kg3 Rf1 Kg4 Rd1 Kf5 Rxd3 Rb2 e4 Rc2 Kd5 Kf4 b3 Re2 c5
Rb2 c4 Kf5 Kd4 Ke6 c3 Rxb3 c2 Rb7 c1=Q Rd7 Ke3 Re7 Ke2 Rf7 Rf3 Rh7)
{-12.05/28 28} Rf3 62. Rd2 (Rd2 b5 Kg2 Rf5 Rc2 Kb4 Rc1 Kxa4 Ra1 Kb4 Kh3 a4
Kg4 Rf4 Kg5 a3 d4 Rxd4 Kf6 e4 Rb1 Kc3 Rc1 Kb2 Re1 a2 Re2 Kb3 Re3 Ka4 Re1 c5
Rc1 e3 Re1 e2 Kf7 Kb3 Kg8 c4 Rxe2 a1=Q Re6 c3 Rg6 c2 Rg3 Kb2 Rg6)
{-15.00/29 33} Rf4 63. Rc2+ (Rc2 Kd4 Rb2 Kxd3 Rxb7 Rxa4 Kg3 Rc4 Ra7 a4 Kf3
Rf4 Kg2 Kc2 Kg3 Rd4 Kg2 Kb3 Kg3 a3 Rb7 Rb4 Rf7 a2 Rf3 Kc4 Kg4 a1=Q Kg5 Qg1
Kf6 Qd4 Rh3 e4 Ke6 e3 Rg3 e2 Kf5 Rb5 Ke6 e1=Q Re3 Qdxe3 Kd7 Qf3)
{-250.00/35 15} Kd6 64. Rb2 (Rb2 Rb4 Ra2 Kc5 Rc2 Kd5 Kg3 Rxa4 Kf2 Rf4 Ke1
b5 Ra2 a4 Ke2 Rh4 Ra1 Rh2 Kf3 Rh3 Kg4 Rxd3 Kf5 a3 Rb1 e4 Kf4 a2 Ra1 Rf3 Kg4
Ra3 Rd1 Kc4 Rc1 Kb3 Ra1 Kb2 Rf1 a1=Q Rf2 Kb3 Kf5 e3 Rf3 Qg1 Ke5 Kb2 Kd6 e2
Rxa3) {-15.95/31 17} Rb4 65. Ra2 (Ra2 Kc5 Re2 Kd5 Kg3 Rxa4 Kf2 b5 Ke1 Rf4
Kd2 b4 Kc2 a4 Rg2 Rf1 Kb2 Rf3 Kc2 Kd4 Rg4 Rf4 Rg2 c5 Rh2 a3 Rh8 Rf2 Kb3 Rb2
Ka4 a2 Rh1 Rb1 Rh6 a1=Q Kb5 Rb2 Rd6 Ke3 Rb6 Rc2 d4 cxd4 Rg6 d3 Rg4 b3)
{-17.30/32 0} Rb3 66. Rd2 (Rd2 Kd5 Kg2 Rb4 Kf3 Rxa4 Rh2 b5 Rh7 b4 Rb7 Kd4
Rg7 Kxd3 Rg4 Ra2 Rg6 c5 Rd6 Kc4 Ke4 b3 Rd1 b2 Rb1 a4 Kxe5 a3 Re1 Ra1 Re4
Kb5 Kd6 b1=Q Re7 Kc4 Kc7 Qd3 Re8 Qh7 Kc8 Rd1 Rd8 Rxd8 Kxd8) {-14.25/27 21}
Ra3 67. Rb2 (Rb2 Kc7 Rf2 Rxa4 Rf7 Kb6 Kg3 Rd4 Rh7 a4 Kf3 Rxd3 Ke4 Rd5 Rh2
Kc5 Rh7 b5 Ra7 Kd6 Kf5 Rd4 Ra8 Rf4 Kg6 e4 Rd8 Kc5 Rd1 a3 Rc1 Kd5 Rd1 Ke5
Rd2 b4 Rc2 b3 Rxc6 a2 Ra6 Rf1 Ra3 Rg1 Kf7 a1=Q) {-9.43/28 12} Kc7 68. Rg2
(Rg2 Rxa4 Rg7 Kb6 Kg2 Rd4 Kf3 Rxd3 Ke4 Rd5 Rg1 a4 Rb1 Kc7 Rh1 b5 Ra1 Kd6
Kf5 Rd4 Kg6 e4 Kf5 Kd5 Kf6 Rd3 Ke7 a3 Kd8 b4 Kc7 e3 Kb6 e2 Re1 a2 Rxe2 b3
Rh2 a1=Q Rh5 Qe5 Rh1 b2 Ka6 Ra3 Kb7 Qe7 Kb8) {-14.45/30 16} Rxd3 69. Rg7+
(Rg7 Rd7 Rg5 Kd6 Kg3 Rf7 Kg4 Rf4 Kh5 Rxa4 Kg6 Rd4 Kf7 a4 Rg6 Kd5 Rg2 b5 Ra2
e4 Ke7 e3 Kd7 Re4 Re2 a3 Kc7 b4 Rc2 e2 Kb6 e1=Q Rc5 Kd4 Rxc6 b3 Rd6 Kc3 Kc7
a2 Rd7 b2 Kd8 Re8 Kc7 Qe5 Rd6) {-19.02/28 9} Kb6 70. Kg2 (Kg2 Rd4 Kf3 Rxa4
Rg5 e4 Ke3 c5 Kf4 Rd4 Rg2 a4 Ke5 Kc6 Ra2 b5 Ra3 Rd3 Ra2 e3 Rg2 c4 Rg6 Kc5
Rg2 a3 Ke4 b4 Rg5 Kc6 Rg6 Kb5 Rg5 Ka4 Re5 a2 Re8 e2 Kf5 b3 Rxe2 a1=Q Re7 b2
Rc7 Qf1 Ke6 Re3 Kd7 Qf5 Kd8) {-16.72/30 16} Rd4 71. Kf3 (Rxa4 Re7 Rd4 Rxe5
Rd5 Re1 Kc5 Rc1 Kd6 Ra1 b5 Rxa5 b4 Ra1 b3 Kc5 Rc1 Kb4 Rb1 b2 c5 c4 Rf1 b1=Q
Rf4 Rd3 Rxc4) {-16.72/30 27} Rxa4 72. Rg5 (Rg5 Rd4 Rxe5 a4 Ke3 Rd5 Re4 Ka5
Kf2 b5 Kg3 Kb6 Re8 Kc5 Kh4 b4 Rb8 a3 Re8 b3 Rb8 a2 Ra8 b2 Rxa2 b1=Q Ra5 Kd6
Rxd5) {-250.00/44 2} e4+ 73. Ke3 (Ke3 c5 Rg6 Kb5 Rg7 b6 Kf4 Rd4 Ke5 a4 Rg2
Kc4 Rc2 Kb3 Rc1 a3 Kf6 a2 Re1 b5 Ke7 b4 Ke6 Kc4 Ke7 b3 Ra1 b2 Rxa2 b1=Q Rh2
e3 Rh5 Qg6 Rxc5 Kxc5) {-250.00/37 10} Rc4 74. Rg6 (Rg6 a4 Rg8 c5 Ra8 Rd4
Kf4 Kb5 Rg8 Kc4 Kf5 b5 Ra8 Kb3 Ra7 a3 Rg7 a2 Rg1 e3 Kg6 e2 Re1 Rd1 Rxe2
a1=Q Re3 Qc3 Rxc3 Kxc3) {-250.00/43 13} a4 75. Rg8 (Rg8 Kb5 Ra8 Kb4 Rb8 a3
Rxb7 Ka4 Ra7 Kb3 Ra8 a2 Rb8 Rb4 Ra8 Ra4 Rb8 Kc4 Rc8 c5 Kf4 a1=Q Kf5 Ra6
Rxc5 Kxc5) {-250.00/37 2} Kb5 76. Rg5+ (Rg5 Kb4 Rg1 a3 Ra1 Kb3 Rb1 Kc2 Rxb7
a2 Ra7 Kb3 Kf4 Rc1 Rb7 Kc4 Rb2 a1=Q Re2 Qd4 Rxe4 Qxe4) {-250.00/44 6} Kb4
77. Rg7 (Rg7 a3 Rxb7 Kc3 Ra7 Kb3 Ra6 a2 Kf2 e3 Ke2 Ra4 Rb6 Ka3 Rxc6 a1=Q
Rd6 Qd4 Rd5 Qxd5) {-250.00/42 4} b5 78. Rg1 (Rg1 a3 Kf4 a2 Re1 Kb3 Rh1 Kb2
Rh2 Rc2 Rxc2 Kxc2 Ke3 c5 Kxe4) {-250.00/42 9} a3 79. Kf4 (Kf4 a2 Re1 Kb3
Kf5 Ra4 Re3 Kc2 Re1 a1=Q Rxa1 Rxa1 Kxe4) {-250.00/46 3} a2 80. Ke5 (Ke5 Kb3
Kd6 Kb2 Rg2 Rc2 Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/47 3} Kb3 81. Kd6 (Kd6
Kb2 Rg2 Rc2 Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/49 2} Kb2 82. Rg2+ (Rg2 Rc2
Rxc2 Kxc2 Kd7 a1=Q Kxc6) {-250.00/53 4} Rc2 83. Rxc2+ (Rxc2 Kxc2 Kd7 a1=Q
Kxc6) {-250.00/70 47} Kxc2 84. Kd7 (Kd7 a1=Q Kxc6) {-250.00/70 0} a1=Q 85.
Kxc6 {-250.00/70 0} Qe5 0-1[/pgn]

[/pgn]

[lkaufman]

Two draw, one loose. Interesting games. What do you think about this result Larry ?

Well, I was hoping for an even score, which doesn't look likely now. But this is really a huge handicap. After 172 self-play games of new Komodo giving itself this two pawn handicap at 15' + 5" on two threads, the score is zero wins, two draws, and 170 losses! That's a 901 elo gap. The elo value of a handicap rises with the time control, so at 45' + 15" it would probably exceed one thousand elo points. But of course there are various factors that might make this number either too low or too high for a human match.

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

On such large a span, the rating model becomes important. FIDE uses Gaussian instead of logistic used in all engine ratings. Gaussian ELO difference for 0.994 is 700 instead of 900 for logistic. Then, the error margins (2SD) for your result after 170 games are about 160 ELO points lower bound and even larger higher bound, similar to a result after 10 games of an even match. I first matched time control odds to get fairly equal strength, then let them play a balanced match. No large ELO differences, but I had to assume that I know the value ELO (logistic or Gaussian, doesn't matter for small intervals) of doubling time control from ultra-fast to long TC, connected doubling by doubling, but I did derive that 2 years ago or so. I agree that the handicap can be larger than 600-650 ELO points I derived, but I doubt it is larger than 700 ELO points of connected small intervals, i.e. maybe Gaussian points or maybe logistic. Once I found for several engines that they obey rather a Gaussian than a logistic, but it's not established yet.

Good points. I've wondered whether Logistic or Gaussian was a better fit for engine results; if it's Gaussian as you say then that is part of the explanation. My final result after 200 games was 199 to 1 (two draws) which is 928 elo using Logistic but quite a bit less using Gaussian. Actually I have some data that supports your claim, namely that I ran the same test at one pawn (f2) handicap and got only minus 337 elo logistic. The two pawn handicap is at most double the f2 handicap (based on several criteria) so this does look Gaussian rather than Logistic.
But regarding the connected doublings, the one solid data point I have is that for Komodo, roughly version 1 thru 5, double speed = 90 elo at ccrl and cegt blitz level. This was easy to determine, because for those versions 32 bit ran almost exactly half as fast as 64 bit, so I only needed to compare 32 bit with 64 bit. Of course you are running faster tests than this, but I suppose I could run at 3' + 1" and expect doublings to be worth perhaps 75 elo or so, since Komodo is now much stronger than then. How did you determine the proper doubling value? I think it is lower for most other engines than Komodo. But each doubling is worth more and more at lower levels, so I think it will be hard to get accurate numbers for this.

[Laskos]

Two draw, one loose. Interesting games. What do you think about this result Larry ?

Well, I was hoping for an even score, which doesn't look likely now. But this is really a huge handicap. After 172 self-play games of new Komodo giving itself this two pawn handicap at 15' + 5" on two threads, the score is zero wins, two draws, and 170 losses! That's a 901 elo gap. The elo value of a handicap rises with the time control, so at 45' + 15" it would probably exceed one thousand elo points. But of course there are various factors that might make this number either too low or too high for a human match.

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

On such large a span, the rating model becomes important. FIDE uses Gaussian instead of logistic used in all engine ratings. Gaussian ELO difference for 0.994 is 700 instead of 900 for logistic. Then, the error margins (2SD) for your result after 170 games are about 160 ELO points lower bound and even larger higher bound, similar to a result after 10 games of an even match. I first matched time control odds to get fairly equal strength, then let them play a balanced match. No large ELO differences, but I had to assume that I know the value ELO (logistic or Gaussian, doesn't matter for small intervals) of doubling time control from ultra-fast to long TC, connected doubling by doubling, but I did derive that 2 years ago or so. I agree that the handicap can be larger than 600-650 ELO points I derived, but I doubt it is larger than 700 ELO points of connected small intervals, i.e. maybe Gaussian points or maybe logistic. Once I found for several engines that they obey rather a Gaussian than a logistic, but it's not established yet.

Good points. I've wondered whether Logistic or Gaussian was a better fit for engine results; if it's Gaussian as you say then that is part of the explanation. My final result after 200 games was 199 to 1 (two draws) which is 928 elo using Logistic but quite a bit less using Gaussian. Actually I have some data that supports your claim, namely that I ran the same test at one pawn (f2) handicap and got only minus 337 elo logistic. The two pawn handicap is at most double the f2 handicap (based on several criteria) so this does look Gaussian rather than Logistic.
But regarding the connected doublings, the one solid data point I have is that for Komodo, roughly version 1 thru 5, double speed = 90 elo at ccrl and cegt blitz level. This was easy to determine, because for those versions 32 bit ran almost exactly half as fast as 64 bit, so I only needed to compare 32 bit with 64 bit. Of course you are running faster tests than this, but I suppose I could run at 3' + 1" and expect doublings to be worth perhaps 75 elo or so, since Komodo is now much stronger than then. How did you determine the proper doubling value? I think it is lower for most other engines than Komodo. But each doubling is worth more and more at lower levels, so I think it will be hard to get accurate numbers for this.

I give here links to my limited "studies" on these issues. The thread about Gaussian versus logistic for several engines:
http://www.talkchess.com/forum/viewtopi ... 66&start=0

And the thread about ELO gain from doubling of nodes in self-play of Houdini, from very small values to some blitz level:
http://www.talkchess.com/forum/viewtopi ... 33&start=0

I found for c2 and f2 pawns handicap that x100 time handicap or 2700+15'' versus 27+0.15'' is a good match, and in 16 games on one core I got +4 -3 =9 result for time handicapped engine, meaning from the plot at the end of the second thread about 600-650 ELO points. The error margins are still pretty high, so 700 points difference is still possible, but this derivation is not affected by distortion due to rating model, because it's based on connected smaller differences.

[lkaufman]

Two draw, one loose. Interesting games. What do you think about this result Larry ?

Well, I was hoping for an even score, which doesn't look likely now. But this is really a huge handicap. After 172 self-play games of new Komodo giving itself this two pawn handicap at 15' + 5" on two threads, the score is zero wins, two draws, and 170 losses! That's a 901 elo gap. The elo value of a handicap rises with the time control, so at 45' + 15" it would probably exceed one thousand elo points. But of course there are various factors that might make this number either too low or too high for a human match.

I am quite puzzled as to why Komodo can give this handicap to some very strong unrelated engines in blitz while it does so much worse in self-play. There are several factors at work here, but the magnitude of the difference is hard to explain.

On such large a span, the rating model becomes important. FIDE uses Gaussian instead of logistic used in all engine ratings. Gaussian ELO difference for 0.994 is 700 instead of 900 for logistic. Then, the error margins (2SD) for your result after 170 games are about 160 ELO points lower bound and even larger higher bound, similar to a result after 10 games of an even match. I first matched time control odds to get fairly equal strength, then let them play a balanced match. No large ELO differences, but I had to assume that I know the value ELO (logistic or Gaussian, doesn't matter for small intervals) of doubling time control from ultra-fast to long TC, connected doubling by doubling, but I did derive that 2 years ago or so. I agree that the handicap can be larger than 600-650 ELO points I derived, but I doubt it is larger than 700 ELO points of connected small intervals, i.e. maybe Gaussian points or maybe logistic. Once I found for several engines that they obey rather a Gaussian than a logistic, but it's not established yet.

Good points. I've wondered whether Logistic or Gaussian was a better fit for engine results; if it's Gaussian as you say then that is part of the explanation. My final result after 200 games was 199 to 1 (two draws) which is 928 elo using Logistic but quite a bit less using Gaussian. Actually I have some data that supports your claim, namely that I ran the same test at one pawn (f2) handicap and got only minus 337 elo logistic. The two pawn handicap is at most double the f2 handicap (based on several criteria) so this does look Gaussian rather than Logistic.
But regarding the connected doublings, the one solid data point I have is that for Komodo, roughly version 1 thru 5, double speed = 90 elo at ccrl and cegt blitz level. This was easy to determine, because for those versions 32 bit ran almost exactly half as fast as 64 bit, so I only needed to compare 32 bit with 64 bit. Of course you are running faster tests than this, but I suppose I could run at 3' + 1" and expect doublings to be worth perhaps 75 elo or so, since Komodo is now much stronger than then. How did you determine the proper doubling value? I think it is lower for most other engines than Komodo. But each doubling is worth more and more at lower levels, so I think it will be hard to get accurate numbers for this.

I give here links to my limited "studies" on these issues. The thread about Gaussian versus logistic for several engines:
http://www.talkchess.com/forum/viewtopi ... 66&start=0

And the thread about ELO gain from doubling of nodes in self-play of Houdini, from very small values to some blitz level:
http://www.talkchess.com/forum/viewtopi ... 33&start=0

I found for c2 and f2 pawns handicap that x100 time handicap or 2700+15'' versus 27+0.15'' is a good match, and in 16 games on one core I got +4 -3 =9 result for time handicapped engine, meaning from the plot at the end of the second thread about 600-650 ELO points. The error margins are still pretty high, so 700 points difference is still possible, but this derivation is not affected by distortion due to rating model, because it's based on connected smaller differences.

The problem is that you are using Houdini to determine the values. I'm pretty sure that with Komodo the doubling values are substantially higher, maybe 25% or so. So 600-650 could easily be 800. You have to use the same (or at least similar) engine for the match and for the doublings for it to be valid. Houdini was a monster at hyperspeed chess (game in a few seconds), not so great at normal blitz. Komodo is the opposite.